Quick viewing(Text Mode)

Conference Digest

Conference Digest

2012 IEEE International Conference on and (ROBIO 2012)

Conference Digest

December 11-14, 2012 ,

Sponsored by IEEE Robotics and Society (RAS) South China University of Technology (SCUT) National Natural Science Foundation of China (NSFC) State Key Laboratory of Robotics and System, HIT IEEE Systems, Man, and Society (SMC) Robotics Society of Japan (RSJ) Japan Society of Mechanical Engineering (JSME) Mechanical Engineering Society (GDMES) ROBIO 2012 PROCEEDINGS

Additional copies may be ordered from:

IEEE Sevice Center 445 Hoes Lane P.O. Box 1331 Piscataway, NJ 08855-1331 USA.

IEEE Catalog Number: CFP12581-ART ISBN: 978-1-4673-2127-3 IEEE Catalog Number (USB drive): CFP12581-USB

Copyright and Reprint Permission:

Abstracting is permitted with credit to the source. Libraries are permitted to pho- tocopy beyond the limit of U.S. copyright law for private use of patrons those articles in this volume that carry a code at the bottom of the first page, provided the per-copy fee indicated in the code is paid through Copyright Clearance Center, 222 Rosewood Drive, Danvers, MA 01923. For other copying, reprint or republication permission, write to IEEE Copyrights Manager, IEEE Operations Center, 445 Hoes Lane, P.P. Box 1331, Piscataway, NJ 08855-1331. All rights reserved. Copyright ©2012 by the Institute of Electrical and Electronics Engineers, Inc.

The Institute of Electrical and Electronics Engineers, Inc Welcome Message On behalf of the IEEE ROBIO 2012 Organizing Committee, we would like to extend a warm welcome to you all to the 2012 IEEE International Conference on Robotics and Biomimetics. ROBIO is a maturing and vibrant international conference running an- nually since 2004 and gaining increasing international prominence especially in areas of biomimicking and biological inspired robotics research. We are pleased to bring you to the 2012 edition as a forum that facilitates timely and productive academic and scientific exchanges among researchers in a wide range of research topics in robotics, biomimetics and beyond. A total 553 papers were submitted to this year’s conference, from 39 countries and regions, with more than 55% from outside Mainland China. Members of the ROBIO Program Committee worked very hard to ensure a fair and careful review process, which is critical in order to maintain the high standard of this conference. After a thorough review process, 369 or about 66% of the submitted papers were accepted for oral presentation at ROBIO 2012. The technical program runs for three days on December 11, 12, and 13, in seven parallel tracks. We are fortunate to be able to anchor the technical program with four plenary talks by leading experts of in robotics and biomimetics: Profs. from United Kingdom, Zexiang Li from Hong Kong, China, Vijay Kumar from the United States, and Koh Hosoda from Japan. Known as Flower City with its pleasant and ever green weather throughout the year, Guangzhou, the host city of ROBIO 2012, is located some 120 km from Hong Kong. Being the third largest Chinese city with a population of 13 million and the host of 2010 Asian Games, Guangzhou plays a pivotal role in the rapid social and economic development of southern China and enjoys a modern transportation infrastructure. Through its social program, ROBIO 2012 allows its attendees to experience and enjoy the Chinese culture, highlighted by a night cruise along which snakes through the city. ROBIO 2012 has been the collective effort of a number of sponsoring societies and organizations, and of our colleagues, friends and volunteers. We would like to take this opportunity to express our sincere gratitude for their help and contributions. The ROBIO conference series is one of the most popular and important events sponsored by IEEE Robotics and Automation Society. In addition, we would like to acknowledge our local sponsoring organization, South China University of Technology. Other members of the organizing committee dedicated a significant amount of their time to ensure this conference a success and, needless to say, we are deeply indebted for their assistance in all areas of the conference. Finally, thank all the authors for your support and participation in ROBIO 2012. This is your conference to be enjoyed and celebrated. We wish you a great conference and a pleasant stay in Guangzhou!

Hong , General Chair Huosheng Hu, Program Chair

Table of Contents

Advisory Board 2 Organizing Committee 3 Program Committee 5 General Information 6 Local Guide to Guangzhou 12 Plenary I: Prof. Kevin Warwick 17 Plenary II: Prof. Zexiang Li 19 Plenary III: Prof. Vijay Kumar 21 Plenary IV: Prof. Koh Hosoda 22 Tutorials 23 Technical Sessions: Tuesday, December 11 27 Technical Sessions: Wednesday, December 12 49 Technical Sessions: Thursday, December 13 71 Index of Session Chairs 93 Index of Authors 95

1 2012 IEEE International Conference on Robotics and Biomimetics

Honorary Chair

Yingjun Wang South China University of Technology, China

Advisory Council

T. J. Tarn Washington University, USA

Toshio Fukuda Nagoya University, Japan

ROBIO Steering Committee

Wen J. Li, City University of Hong Kong, China Yunhui Liu, Chinese University of Hong Kong, China Shugen Ma, Ritsumeikan University, Japan Max Q. H. Meng, Chinese University of Hong Kong, China Ning Xi, City University of Hong Kong, China Hong Zhang, University of Alberta, Canada

2 Organizing Committee

General Chair Hong Zhang, University of Alberta, Canada

General Co-Chairs Xianmin Zhang, South China University of Technology, China Peter Corke, Queensland University of Technology, Australia Gregory Dudek, McGill University, Canada Kazuhiro Kosuge, Tohoku University, Japan

Program Chair Huosheng Hu, University of Essex, UK

Program Co-Chairs Yisheng Guan, South China University of Technology, China Hajime Asama, University of Tokyo, Japan Tim Lueth, Technical University of Munich, Germany Eddie Tunstel, Johns Hopkins University, USA

Awards Chairs Wei-Hsin Liao, Chinese University of Hong Kong, China Martin Adams, University of Chile, Chile Yili Fu, Harbin Institute of Technology, China

Tutorial/Workshop Chairs Minoru Hashimoto, Shinshu University, Japan Kazuhito Yokoi, ISRI/AIST, Japan Jianwei Zhang, University of Hamburg, Germany

Organized Sessions Chairs Shuxiang Guo, Kagawa University, Japan Timothy Barfoot, University of Toronto, Canada Abderrahmane Kheddar, LIRMM, France

3 Publicity Chairs I-Ming Chen, Nanyang Technology University, Singapore Chris Smith, Gonzaga University, USA

Publication Chairs Zhidong Wang, Chiba Institute of Technology, Japan Lizhen Wang, Yunnan University, China

Local Arrangement Chair Nianfeng Wang, South China University of Technology, China

Finance Chairs Melanie Calvert, University of Alberta, Canada Wei Sun, South China University of Technology, China Hongxia Yang, South China University of Technology, China

4 Program Committee

Chair: Huosheng Hu Co-Chairs: Yisheng Guan, Hajima Asama, Tim Lueth, Eddie Tustel

Members (in alphabetic order) Adams, Martin Kurisu, Masamitsu Shen, Yantao Aphiratsakun, Narong Lai, King Suebsomran, Anan Arisumi, Hitoshi Li, Dayou Sun, Dong Azzouzi, Messaouda Li, Peng Suthakorn, Jackrit Bayraktaroglu, Zeki Y. Li, Wen Jung Suzuki, Tatsuya Bellotto, Nicola Li, Yangmin Tahara, Kenji Bowling, Alan Li, Youfu Tan, Jindong Capi, Genci Liao, Wei-Hsin Tan, Min Carey, Stephanie Limsakul, Chusak Tarn, T.J. Chen, Ling Liu, Jinguo Tatsuno, Kyoichi Chen, Wenhua Liu, Honghai Ting, Yung Chen, Xinkai Liu, Yong Theodoridis, Theo Duan, Xingguang Liu, Yunhui Tsuji, Toshio Deelertpaiboon, Chirdpong Lu, Zhenli Umeda, Kazunori Elhajj, Imad Luo, jun Wang, Han Fu, Yili Ma, Lei Wang, Hesheng Fukuda, Osamu Ma, Shugen Wang, Kundong Fung, Wai-keung Meng, Max Wang, Nianfeng Gu, Dongbing Meng, Qinggang Wang, Zhidong Gu, Jason Meng, Qinghao Whidborne, James Guo, Shuxiang Menon, Carlo Xiao, Jizhong Guo, Yi Miao, Lei Xu, De Hasegawa, Yasuhisa Ming, Aiguo Yagi, Yasushi Harada, Kensuke Mir-Nasiri, Nazim Yamashita, Atsushi Hirai, Shinichi Mitani, Atsushi Yano, Kenichi Hirata, Yasuhisa Nakajima, Shuro Yu, Yong Huang, Jian Neubert, Jeremiah Yuan, Kui Huang, Qiang Ogawara, Koichi Yue, Yong Huang, Xinhan Oscar, Ylaya Chuy Zhang, Jianwei Inoue, Kenji Oskoei, Mohammadreza Zhang, Xianmin Ito, Kazuyuki Ota, Yusuke Zhang, Yi Iwashita, Yumi Oyekan, John Zhou, Yu Jia, Songmin Pan, Wei Zhou, Huiyu Kawabata, Kuniaki Peng, Jun Zhu, Xiangyang Kruusmaa, Maarja Roning, Juha Zhu, Xiaorui Kuno, Yoshinori Ryuh, Yongsun Zhuang, Yan

5 General Information

Conference Registration

The conference registration desk, located in the grand lobby on the first floor of Crowne Plaza City Centre, will be open according to the schedule below for on-site registration as well as for the pre-registration attendees to pick up their registration package.

Day Monday, Dec. 10 Tuesday, Dec. 11 Wednesday, Dec. 12 Thursday, Dec. 13

Time 14:00 - 18:30 08:30 - 16:30 08:30 - 16:30 08:30 - 16:30

Four-day Conference Schedule

The conference technical sessions and social functions cover a four-day period, Decem- ber 11-14, from Tuesday to Friday. The overall schedule is shown in the table below.

Time/Day Tuesday, Dec. 11 Wednesday, Dec. 12 Thursday, Dec. 13 Friday, Dec. 14

Plenary Talk I Plenary Talk II Plenary Talk IV Morning Parallel Sessions Parallel Sessions Parallel Sessions Tutorials (Tu1) (We1) (Th1)

Parallel Sessions Parallel Sessions Parallel Sessions Afternoon (Tu2) (We2) (Th2) Parallel Sessions Parallel Sessions Parallel Sessions (Tu3) (We3) (Th3)

Evening Open Plenary Talk III Pearl River Cruise Award Banquet (Farewell Party)

Four plenary talks will be given in Crystal Ballroom (Plenary talks by Professor Zexi- ang Li, Vijay Kumar and Koh Hosoda will be in Crystal Ballroom 1, while that by Kevin Warwick in Crystal Ballroom 2) on the third floor of the Crowne Plaza City Centre (please refer to Figure 1). The seven parallel technical sessions will be held in rooms Jade 1, Jade 2, Jade 3, Jade 5, Jade 7, Jade 8 and Jade 9, respectively. A more detailed schedule can be found in the table of Conference Program on page 8.

Official Language

The official language of the conference is English. All presentation including discussion

6 Parallel Session Jade Rooms Jade Jade Jade (䪏⷇ख़) Room 7 Room 8 Room 9

Jade Room 3

Jade Jade Jade Elevators Room 1 Room 2 Room Plenary Talk 5 Crystal Ballrooms (∈᱊ख़)

Crystal Crystal Ballroom 1 Ballroom 2

Figure 1: Room layout on Floor 3, Crowne Plaza City Centre and submissions shall be made in English.

Support Facilities

There will be a data projector, a pointer, a microphone and a computer available in all session rooms. The presenters are allowed to use their own computers. Anything else required must be requested by the authors.

Internet Service

Wireless Internet access will be available in the session rooms on the 3th floor, Decem- ber 11-14 at 09:00-18:00.

7 IEEE ROBIO 2012 Conference Program December 11-14, Crown Plaza City Centre, Guangzhou, China

Room Jade 1 Jade 2 Jade 3 Jade 8 Jade 5 Jade 9 Jade 7

Monday, December 10, 2012 18:30- Welcome Reception Crystal Ballroom 2 20:00 Tuesday, December 11, 2012 08:30- Crystal Ballroom 2 Opening Ceremony 09:00 09:00- Plenary Talk 1: TheHuman-RobotMerger Crystal Ballroom 2 10:00 Professor Kevin Warwick, University of Reading, United Kingdom Coffee/Tea Break Tu1-1: Tu1-2: Tu1-3: Tu1-4: Tu1-5: Tu1-6: Tu1-7: 10:20- Humanoid I Legged Robots I Vision/Image Biologically Inspired Grasping and Smart Structures, Service Robots I 11:50 Processing I Robotics I Manipulation I Materials, Actuators I Lunch Tu2-1: Tu2-2: Tu2-3: Tu2-4: Tu2-5: Tu2-6: Tu2-7: 13:30- Humanoid Robots Legged Robots II Vision/Image Biologically Inspired Grasping and Machine Learning I SLAM I 15:00 II Processing II Robotics II Manipulation II Coffee/Tea Break Tu3-1: Tu3-2: Tu3-3: Tu3-4: Tu3-5: Tu3-6: Tu3-7: 15:20- Underwater Robots Biomimicking I Medical & Surgery Mobile Robots I Grasping and Machine Learning II SLAM II 17:05 I Robots I Manipulation III

Wednesday, Dec 12, 2012 09:00- Plenary Talk 2: A Geometric Theory of Robotics and Beyond … Crystal Ballroom 1 10:00 Professor Zexiang Li, Hong Kong University of Science and Technology, China Coffee/Tea Break We1-1: We1-2: We1-3: We1-4: We1-5: We1-6: We1-7: 10:20- Underwater Robots Biomimicking II Medical & Surgery Mobile Robots II Aerial robots I Smart Structures, Service Robots II 11:50 II Robots II Materials, Actuators II Lunch We2-1: We2-2: We2-3: We2-4: We2-5: We2-6: We2-7: 13:30- Micro/Nano Biomimicking III Medical & Surgery Human- Aerial robots II Smart Structures, Service Robots III 15:00 Devices & Robots Robots III Interaction I Materials, Actuators III Coffee/Tea Break We3-1: We3-2: We3-3: We3-4: We3-5: We3-6: We3-7: 15:20- Intelligent Systems Biologically Sensor-Networks Human-Robot Rehabilitation Machine Learning III Man-Machine 16:50 I Inspired Robotics Interaction II Robots I Interface I 17:00- Plenary Talk 3: Robots that Fly Crystal Ballroom 1 18:00 Professor Vijay Kumar, University of Pennsylvania, USA

18:30- Crystal Ballroom Award Banquet 20:30 (1 and 2)

Thursday, December 13, 2012 09:00- Plenary Talk 4: Compliant Body as a Source of Intelligence Crystal Ballroom 1 10:00 Professor Koh Hosoda, Osaka University, Japan Coffee/Tea Break Th1-1: Th1-2: Th1-3: Th1-4: Th1-5: Th1-6: Th1-7: 10:20- Intelligent Systems Motion Planning Tele-robotics & Multi-robot I Rehabilitation I Man-Machine 11:50 II I Networked Robots Robots II Interface II Lunch Th2-1: Th2-2: Th2-3: Th2-4: Th2-5: Th2-6: Th2-7: 13:30- Intelligent Systems Motion Planning Vision/Image Multi-robot II Rehabilitation Robot Control II Sensors I 15:00 III II Processing III Robots III Coffee/Tea Break Th3-1: Th3-2: Th3-3: Th3-4: Th3-5: Th3-6: Th3-7: 15:20- Modular Robot Motion Planning Vision/Image Multi-robot III Search Control III Sensors II 16:50 Systems III Processing IV Robots 18:30- Pearl River Night Cruise (Farewell Party) 20:30 Friday, December 14, 2012 09:00- Tutorial I: Medical Robotics Jade 9 10:00 Professor Tim C. Lueth, Technical University of Munich, Germany 10:00- Tutorial II: Robotics as Service Enabler Jade 9 11:00 Professor Kazuhiro Kosuge, Tohoku University, Japan Special Colloquium: People Behaving Badly, Robots Behaving Better? Jade 9 11:00- Embedding Ethicsin Autonomous Robotic Systems 12:00 Professor Ron Arkin, Georgia Institute of Technology, USA Social and Cultural Functions

Welcome Reception is held in Crystal Ballroom 2 on the 3rd floor of Crowne Plaza City Centre, on December 10 at 18:30 - 20:00, to welcome participants of ROBIO 2012. Refreshment drinks and light finger food are provided.

Daily lunches, in buffet-style with a mixture of western and Chinese foods, both regular and vegetarian, are provided to ROBIO attendees in Fresh Restaurant on the 5th Floor of Crowne Plaza, on December 11-13 at 12:00 - 13:30.

Award Banquet is held in Crystal Ballroom (1 and 2), on the 3rd floor of Crowne Plaza City Centre, on December 12 at 18:30 - 20:30. Dinner will be first served, followed by presentation of the finalists and winners of the best paper awards in four categories and a brief introduction of ROBIO 2013.

Pearl River Night Cruise (Farewell Party) is organized, on December 13 at 18:00 - 20:30, where ROBIO 2012 participants can relax and take in the views of Guangzhou at night with its landmark buildings and bridges decorated in spectacular lights while enjoying a buffet-style dinner on the boat. Please wait at the front entrance of Crowne Plaza at 18:00, to board the buses that take us to the boat dock.

Conference Venue

ROBIO 2012 is held in Crowne Plaza City Centre, located in the center of Guangzhou City. The hotel and its address are shown in Figure 2. The conference rooms are on the 3rd floor of the hotel, whose layout is shown in Figure 1.

Transportation The hotel can be reached from the Guangzhou Baiyun International Airport by bus, metro or taxi. An airport shuttle (Line 2A) takes you directly to the hotel, at a cost of 18 RMB in about 45 minutes. One-way taxi from the airport to the hotel costs approximately 100 RMB. One can also reach the hotel from the airport by metro, boarding at Airport South (Å|H), changing to Line 5 at (¾ô#¢) station, and getting off at Taojin (=7) station. Then using Exit B, the hotel is approximately 400 meters away (a 5-minute walk), please refer to Figure 3. One way metro fare from the airport to the hotel is 8 RMB. Finally, if you arrive in Guangzhou by train at one of the three train stations (Guangzhou Railway Station, Guangzhou East Station, and Guangzhou South Station), metro service is also available (from Line 2 or 3 to Line 5), please refer to Figure 4 for all existing metro lines. As well, it is possible to arrange limousine service with the conference hotel for pick up or drop off from the Baiyun International Airport, at 500 RMB per trip with 24 hour advance notice. Please contact the hotel directly for its limousine service (reserva- [email protected]). If you travel to Guangzhou through Hong Kong International Airport, one convenient option to reach the conference hotel is direct bus service from Hong Kong International Airport to Garden Hotel, which is located at Taojin Metro station.

9 Crowne Plaza City Centre (¥%•)bFËA) Address: 339 Huanshi Road East, , Guangzhou £2²½D«‚½À´339Ò¤ Tel: 86-20-83638888, Fax: 86-20-83636666

Figure 2: Crowne Plaza City Centre (the conference venue)

Crowne Plaza City Centre

~ 400 m

The

Exit B

Figure 3: From Metro Station (Taojin) to Crowne Plaza Hotel

10 Airport South

Guangzhou Station

Guangzhou Taojin East Station

Guangzhou South Station

Figure 4: Metro lines in Guangzhou

11 Local Guide to Guangzhou

The host city of the conference, Guangzhou (z², 2²), is located at the central- south of Guangdong Province, China, north of the Pearl River Delta, at the entrance of Pearl River (Zhu Jiang) to the Pacific Ocean, as shown in Figure 5. The Pearl River, the third largest river of China, runs through the city and is navigable to the South China Sea. Situated in such an excellent geographical region, Guangzhou is called China’s South Gate.

Guangzhou

Shenzhen Hong Kong

Macau

Figure 5: Location of Guangzhou in Guangdong Province

Guangzhou is a prosperous metropolis full of vigor. It is the capital city of Guangdong, and the center of its political, economic, scientific, educational and cultural life. Sui (Ears of Rice) is short for Guangzhou and the City of Ram (Yang Cheng) is also an alias of Guangzhou. The kapok is the city flower and the kapok tree is the city tree of Guangzhou. Covering an area of 7434.4 square kilometers (2870 square miles), Guangzhou has a population of 12.78 million as of 2010, including a 3.7 million transitory population. With the opening of China to the outside world, a large number of people from other regions of China swarmed into Guangzhou, one of the first “open” cities in China. This has accelerated its economic development. Guangzhou is a famous historical city. Its recorded history begins with China’s con- quering of the area during the Qin Dynasty. Its earliest recorded name Panyu expanded when it became capital of the Nanyue Kingdom (H) in 206 BC, the territory of the Nanyue Kingdom included what is now Vietnam. In ancient days, Guangzhou was the capital city for three Chinese dynasties: the NanYue (South Yue), the Nanhan (South

12 Han) and the Nanming (South Ming). Thus it was put in the list of the 24 most fa- mous historical cultural cities and became a tourist destination. You cannot understand most Chinese cities deeply until you know their history. This is true of Guangzhou. Many historic sights such as the Museum of the Western Han Nanyue King’s Mausoleum, the Zhenhai Tower, and the Chan Clan Ancestral Hall tell us the 2,000-year history of Guangzhou. Guangzhou is famous as a hometown for overseas Chinese. It boasts the largest pop- ulation of overseas Chinese people. These overseas Chinese do a great deal of good for Guangzhou: opening international markets, bridging Guangzhou and the rest of the world, and establishing many schools, hospitals, nurseries, kindergartens and rest homes in Guangzhou. Hotel facilities and service in Guangzhou are fine. A word of warning: Do make a reservation beforehand if you decide to visit Guangzhou during the Canton Fair. For visitors, especially those plan to attend the Canton Fair, transportation in Guangzhou is more and more convenient. You can choose plane, train, ship or bus.

Climate

Located just south of the Tropic of Cancer, Guangzhou has a humid subtropical cli- mate influenced by the Asian monsoon. Summers are wet with high temperatures, high humidity and a high heat index. Winters are mild and comparatively dry. Guangzhou has a lengthy monsoon season, spanning from April through September. Monthly averages range from 13.6◦C in January to 28.6◦C in July, while the annual average temperature is 22.6◦CC. December is one of the coolest months of the year, with temperature varying from several Celsius degrees to more than ten Celsius degrees. It does not snow all over the year.

Business Hours

Most offices are open from 9:00 to 17:00 on weekdays except for public holidays. S- tores and restaurants are open during 8:00-22:00 everyday. Most banks and post offices are open on weekend, but closed earlier.

Local Time

GMT +8 hours, Beijing time (BJT).

Electricity

The electric system is 220 Volt AC (50Hz). There are many plugs and sockets avail- able in the hotel. Travelers with shavers, computers and other personal electronics should carry a plug-adapter kit if necessary.

Culture

Cantonese culture is one important part of Chinese culture, with its own special fea- tures in many aspects such as language, Cantonese cuisine, Cantonese opera,

13 Cantonese people, Guangdong music (genre), , Guangzhou Sym- phony Orchestra, and so on. Cantonese language is the first language for half of the 13 million residents of the provincial capital Guangzhou, while the other half speak mainly Mandarin. Other languages such as Hakka are spoken in significant numbers as well.

Shopping

Guangzhou retains its ancient customs, but is also a large city full of vigor and cur- rent fashions. Local products include Canton Sculpture, Cantonese Enamel, Cantonese Embroidery, and Canton Bacon. You can find something worth taking home in Shangxia Jiu Road (þeÊ´), Dishifu Road (1›7´), Beijing Road (®´), 101 Dynamics, China Plaza, Jiangnanxi, Liwan Plaza (t 2|), Teem Plaza (Uà¢2|), (‘õ|2|), Wanguo Plaza (I2|), and Zhengjia Square (Grandview Mall, Z2|).

Eating in Guangzhou

Among various things, Cantonese Cuisine (Yue Cai) is perhaps what Guangzhou is most famous for. Trying Cantonese Cuisine, one of eight Chinese famous cuisines, is definitely a must in Guangzhou! Eat delicious food with ingredients you have never heard of. Cantonese Cuisine includes sauces and condiments, dried and preserved ingredients, traditional dishes, deep fried dishes, slow-cooked soup, sea food, noodle dishes, Siu mei, Lou mei, Siu laap, little pan rice, banquet/dinner dishes, and dessert. For your convenience during the conference, general food may be found in some restau- rants surrounding Crowne Plaza City Centre, including Wangpin Stake (¬‡Úh) and Xi’ai Xuan(UOZ). On the opposite side of the Huanshi East Road, there is JUSCO in the Poly Zhonghuan Plaza (|¥‚2|) where approximate 10 restaurants can be found inside. Mini bars of different styles locate on the sides of Taojin Road (=7´).

Tourist Attractions

As an ancient city with long history, Guangzhou has many tourist attractions in- cluding Museum of the Nanyue King’s Mausoleum (ÜÇHÆÔ,), (Ü '), Chen Clan Academy/Chan Clan Ancestral Hall (•¼Ö/•[), Is- land/Shameen Island (â¡), Guangdong Provincial Museum (2ÀŽÆÔ,), Temple of the Six Banyan Trees (8´m), Sacred Heart Cathedral/Stone House (’%Œ,/œ ¿), (~’m), Temple of Bright Filial Piety (1m), Chime-Long Paradise (•ž•W•.), Chime-Long Water Park (•žYþW ), and Sun Yat-sen Memorial Hall (¥ìVg,).

Eight Sights

The Eight Sights of Guangzhou are the eight most famous tourist attractions listed by rulers in different ages. The followings are those chosen through public appraisal and brought out in 2011. ˆ Tayao Xincheng (©ˆ#¢): , Chigang Pagoda, Haixinsha, Flower

14 City Square, Twin Towers, CITIC Plaza and etc. ˆ Zhushui Liuguang (¾Y61): Pearl River (segment from Bai’etan to ), Wharves, Bai’etan, Shamian Island, Yanjiang Lu, , Haixinsha, Guangzhou Convention and Exhibition Center and etc. ˆ Yunshan Diecui ( ìU}): Baiyun Mountain, Yuntai Garden and etc. ˆ Yuexiu Fenghua (Dºu): Yuexiu Mountain, Zhenhai Tower, Five Rams Statue, Sun Yat-sen Memorial Monument and etc. ˆ Guci Liufang (6•): Chen Clan Academy and Guangdong Folk Craftwork Museum. ˆ Liwan Shengjing (t ‘µ): Litchi Bay, Liwan Lake, Arcades on Enning Lu, Xiguan Residences and etc. ˆ Kecheng Jinxiu (‰¢

Pear River Cruise

The Pearl River is the third longest river in China, measuring about 2,129 kilometers in length. The Guangzhou section of the river is where Guangzhou’s most scenic spots are located (see Figure 6). Many famous sights are clustered along the river and are the reasons for many people taking a trip to Guangzhou. The Pearl River night cruise is unique to Guangzhou. The night cruise on the Pearl River is one of the tourist highlights in Guangzhou. The beauty of the Pearl River at night can be compared with that of Hong Kong’s Victoria Park. The cruise section from the White Swan Pond to is the highlight of the Pearl River cruise. Scenery along this part is so fascinating, with jade-coloured water and coloured lights on shore (refer to Figure 7). Sights on Pearl River Night Cruise include White Swan Pond (Bai’etan), Shamian Building Complex, Aiqun Mansion, Tower Shadow Building, Riverside Bar Corridor, Ersha Island, Haixinsha, and Guangzhou Tower. You can cruise on a luxury boat along the Pearl River and enjoy the enchanting land- scapes on both banks. You can stand on the deck and feel the breeze of the river on your face and hear the sound of water splashing against the side of the ship. The cruise is immensely popular among local residents and tourists alike.

Emergency Help

Alarm Call: 110 (toll free); Emergency Medical Call: 120 (toll free).

15 Figure 6: The Pearl River in a clear autumn day

Figure 7: Night view of the Pearl River

16 Plenary Talk I

Tuesday, December 11, 09:00 - 10:00 Crystal Ballroom 2 The Human-Robot Merger

Professor Kevin Warwick University of Reading, United Kingdom

Abstract In this presentation a practical look is taken at how the use of implant and electrode technology can be employed to create biological brains for robots, to enable human en- hancement and to diminish the effects of certain neural illnesses. In all cases the end result is to increase the range of abilities of the recipients. An indication is given of a number of areas in which such technology has already had a profound effect, a key element being the need for a clear interface linking a biological brain directly with computer technology. The emphasis is clearly placed on experimental scientific studies that have been and are being undertaken and reported on. The area of focus is notably the need for a biological/technological connection, where a link is made directly with the cerebral cortex and/or nervous system. The presentation will consider the future in which robots have biological, or part-biological, brains and in which neural implants link the human nervous system bi-directionally with technology and the internet. Brief Biography Kevin Warwick is Professor of Cybernetics at the University of Reading, England, where he car- ries out research in artificial intelligence, control, robotics and . Kevin was born in Coventry, UK and left school to join British Telecom, at the age of 16. At 22 he took his first degree at Aston University, followed by a PhD and research post at Imperial College, London. He subsequently held positions at Ox- ford, Newcastle and Warwick Universities before being offered the Chair at Reading, at the age of 33. As well as publishing over 500 research papers, Kevin’s experiments into implant tech- nology led to him being featured as the cover story on the US magazine, “Wired”. Kevin has been awarded higher doctorates (DSc) both by Imperial College and the Czech A- cademy of Sciences, Prague, and received Honorary Doctorates from Aston University, Coventry University, Bradford University, Robert Gordon University, University of Bed- fordshire and Portsmouth University. He was presented with The Future of Health Tech- nology Award in MIT, was made an Honorary Member of the Academy of Sciences, St. Petersburg, and has received The IEE Senior Achievement Medal, the IET Mountbat- ten Medal and in 2011 the Ellison-Cliffe Medal from the Royal Society of Medicine. In 2000 Kevin presented the Royal Institution Christmas Lectures, entitled “The Rise of the

17 Robots”. Kevin’s present research involves the invention of an intelligent deep brain stimulator to counteract the effects of Parkinson Disease tremors. The tremors are predicted and a current signal is applied to stop the tremors before they start - this is to be trialed in human subjects. Another project involves the use of cultured/biological neural networks to drive robots around - the brain of each robot is made of neural tissue. Perhaps Kevin is though best known for his pioneering experiments involving a neuro- surgical implantation into the median nerves of his left arm to link his nervous system directly to a computer to assess the latest technology for use with the disabled. He was successful with the first extra-sensory (ultrasonic) input for a human and with the first purely electronic telegraphic communication experiment between the nervous systems of two humans.

18 Plenary Talk II

Wednesday, December 12, 09:00 - 10:00 Crystal Ballroom 1 A Geometric Theory of Robotics and Beyond

Professor Zexiang Li Hong Kong University of Science and Technology, Hong Kong, China

Abstract Gaspard Monge established the Ecole Polytechnique in 1794, 5 years after the French Revolution, and created the modern engineering education system in France. The Ecole Polytechnique is credited with the effort of applying scientific method, especially Calculus of Newton and Leibniz, to engineering design and analysis. Monge himself taught a course at the Ecole Polytechnique on machine design using his descriptive geometry and calculus. The n-dimensional Euclidean space (Rn), on which Calculus is based, has served as model spaces of modern engineering for the last 200 years. Today, however, many en- gineering problems especially these arising from robotics, control, manufacturing and mechanism design have structures very different from that of Rn. Continuing to rely on approximation and simplification techniques no longer satisfies the demand of today’s engineering. In this talk, I will present an effort, which was initiated by R. Brockett of Harvard about 30 years ago, and continued by the Berkeley group and then my own group at HKUST for the last 20 years to develop a unified theory, using tools from differentiable manifolds and Lie groups, for robotics, mechanism design and manufacturing research. First, using intuitive examples, I will recollect some of the basic concepts of differen- tiable manifolds and their ”engineering” classifications. Then, I will show how problems in robotics, mechanism design and manufacturing research can be modeled using various types of manifolds as their model spaces. Finally, I will highlight how geometric proper- ties of these spaces are being exploited to provide more efficient solutions for optimization problems defined on these spaces. I will describe how results from this research program are being used as basis for found- ing of three companies, one in motion control, one in UAV-based consumer products and the third in assembly automation. Borrowing from L. Page’s words, I credit this effort to“Geometry as inspiration”.

Brief Biography Zexiang Li attended the South-Central University in 1978, received his BS (with hon- or) degrees in and Economics from Carnegie-Mellon University in 1983, his MS degree in EECS in 1985, MA in mathematics and PhD in EECS in 1989, all from the University of California at Berkeley. He worked at ALCOA, the Robotics Institute of CMU and the AI Lab of MIT (89-90). He was an assistant pro- fessor at the Courant Institute of New York University (90-92). In 1992, he joined the Department of Electronic and Computer Engineering of the Hong Kong University

19 of Science and Technology and is currently a professor of the department. He found- ed the Automation Technology Center (ATC) and currently serves as its director.

Zexiang Li received the ALCOA Foundation Fellowship in 1979, and the E. Anthony Fellow- ship in 1983. He was a recipient of the University Scholar award from CMU in 1983, the E.I. Jury award from UC Berkeley in 1989, the Research Initiation award from NSF (US) in 1990, the Out- standing Young Researcher award (Class B) from NSF China in 2000, the LEAD award from AMI, USA in 2001, and the Natural Science award (3rd class) from China in 1997. He became an IEEE Fellow in 2008. Zexiang Li served as a panel member of the Hong Kong Research Grants Council (RGC), an overseas member of the Natural Science Foundation of China (NSFC), and an associate editor for the IEEE Trans. on Robotics and Automation. He was the general Chair for the 2011 IEEE International Conference on Robotics and Automation (ICRA). Zexiang Li’s research areas of interests include multifingered robotic hand, parallel manipulators, workpiece localization and inspection, motion control, precision assembly, and unmanned aerial vehicles (UAVs). He is the author of more than 100 journal and conference papers, and the books A Mathematical Introduction to Robotic Manipulation (CRC Press 1993), and Nonholonomic Motion Planning (Kluwer 1994). Zexiang Li cofounded several companies with his colleagues and students from the Automation Technology Center, including Googol Technology, the first motion control company in China, iFlight Technology (or Dajian Innovation), a leading company in UAV products, Parallex, a company specializing in LED and IC packaging automation equipment, and Lie Group Automation.

20 Plenary Talk III

Wednesday, December 12, 17:00 - 18:00 Crystal Ballroom 1 Robots That Fly

Professor Vijay Kumar University of Pennsylvania, United States

Abstract Autonomous micro aerial robots can operate in three-dimensional unstructured envi- ronments, and offer many opportunities for environmental monitoring, search and rescue, and first response. I will describe the challenges in developing small, agile robots and our recent work in addressing these challenges. I will also discuss the deployment of large numbers of aerial robots, focusing on the control and planning problems with applications to cooperative manipulation and transport, construction, and exploration and mapping.

Brief Biography Vijay Kumar is the UPS Foundation Professor and the Deputy Dean for Education in the School of Engineering and Applied Science at the University of Pennsylvania. He received his Ph.D. in Mechanical Engineering from The O- hio State University in 1987. He has been on the Faculty in the Department of Mechanical Engineering and Applied Mechanics with a secondary appointment in the Depart- ment of Computer and Information Science at the Univer- sity of Pennsylvania since 1987. Dr. Kumar is a Fellow of the American Society of Me- chanical Engineers (ASME) and the Institution of Electri- cal and Electronic Engineers (IEEE). He presently serves on the editorial boards of the IEEE Transactions on Automation Science and Engineering, the ASME Journal of Mechanisms and Robotics and the Springer Stracts in Advanced Robotics (STAR). He is the recipient of the National Science Foundation Presidential Y- oung Investigator award (1991), the Lindback Award for Distinguished Teaching (1996), and best paper awards at DARS 2002, ICRA 2004, ICRA 2008, ICRA 2011, RSS 2009, DARS 2010 and RSS 2011. Most recently he was named the winner of a 2012 IEEE Robotics and Automation Society Distinguished Service Award, the 2012 ASME - nisms and Robotics Committee Award, and a 2012 World Technology Network Award.

21 Plenary Talk IV

Thursday, December 13, 09:00 - 10:00 Crystal Ballroom 1 Compliant Body as a Source of Intelligence

Professor Koh Hosoda Osaka University, Japan Abstract The amazing progress of robots in these decades is supported by the formal compu- tation. The robot and the environment should be modeled precisely, which limits design variation of the robot and possible environment. The desired behavior is also described formally, which limits class of the behavior. On the other hand, animals can behave adaptively in open environments. Their adaptability is currently far superior to that of existing robots. How we can design a robot whose adaptability is comparable to the biological systems? We suppose that the adaptability comes from compliance brought by muscular-skeleton structure and from softness of the skin. Based on this hypothesis, we have built many robots that have soft skin and animal-like muscular skeleton system. I will introduce some of our robots in the talk, and discuss how they achieve adaptive behaviors utilizing their morphology.

Brief Biography Professor Koh Hosoda received his Ph.D. degree in me- chanical engineering from Kyoto University, Japan in 1993. From 1993 to 1997, he was a Research Associate of Me- chanical Engineering Department, Osaka University. From 1997 to 2010, he was an Associate Professor of Graduate School of Engineering, Osaka University. Since 2010, he has been a Professor of Graduate School of Information Science and Technology, Osaka University. In the mean time, From Apr. 1998 to Mar. 1999, he was a guest pro- fessor in Artificial Intelligence Laboratory, University of Zurich.

22 Tutorials

Tutorial I: “Medical Robotics” 09:00 - 10:00, December 14, 2012

Speaker: Prof. Dr.rer.nat. Dipl.-Ing. Tim C. Lueth Institute of Micro Technology and Medical De- vice Technology Technical University of Munich (TUM) Germany

Abstract: Medical robotics did start with Brian Davies first ideas at Imperial College at the end of the 80th for prostate cancer resection. Since the year 2005, robots in medical procedures are considered as helpful tools or assistance for surgery, radiation therapy, and imaging. Companies such as Intuitive Surgical have today a much higher stock market value than robotics companies for industrial applications or service robots. More and more industrial robotics companies drift into this new market. In the talk, different types of surgical assistance systems are described from the past to the current state of the art. In addition a collection of different robot system are presented which are medical products today. Beside the basic principles of these systems, also the procedure to develop and to get a medical approval for such kind of robot systems is explained. At the end of talk, a new concept of single use robot systems is presented for snake-like robots in minimally invasive surgery. Those robots are designed in a half automatic way using selective laser sintering for manufacture the robot kinematic over night.

23 Tutorial II: “Robotics as Service Enabler” 10:00 - 11:00, December 14, 2012

Speaker: Professor Kazuhiro Kosuge Department of Bioengineering and Robotics Tohoku University Japan

Abstract: Robotics research is classified into two types: curiosity-driven robotics research and needs-driven robotics research. Both types of research have created rich foundation for new technological systems. Robotics has become a key technology for the engineering. Many potential applications exist in human environment. Physical human-robot interac- tion is one of key issues for developing a robot in human environments working together with its partner(s). A dance partner robot, which has been developed in our laboratory and dances a waltz with a male dancer, is a research platform for the human-robot in- teraction. How to estimate the following step lead by the male dancer, or how to read the male dancer.s lead, or how to estimate its partner.s intention, was one of the key issues for the development of the robot. In this talk, we will first consider issues relating to physical human-robot interaction and introduce what we have done so far for the dance partner robot. Then, we will consider how the concept of the dance partner robot could be utilized in real world applications. As one of the applications of the human-robot interaction and coordination, PaDY (Parts/tools-Delivery-to-You robot) is introduced. PaDY is a robot, which is being developed for automobile assembly process and assists its partners to assemble parts to a vehicle. Finally an overview of robotics research is introduced and will show how new services will be created by the robotics.

24 Special Colloquium: “People Behaving Badly, Robots Behaving Better? Embedding Ethics in Autonomous Robotic Systems” 11:00 - 12:00, December 14, 2012 Speaker: Professor Ron Arkin Regents’ Professor Georgia Institute of Technology United States

Abstract: Weaponized robotic systems are being introduced into the battlefield at an ever increas- ing pace. The consequences of this technological progress need to be examined carefully. In this talk, I outline the philosophical basis, motivation, theory, and design recommen- dations for the implementation of an ethical control and reasoning system potentially suitable for constraining lethal actions in an autonomous robotic system so that they fall within the bounds prescribed by the Laws of War and Rules of Engagement. It is a fur- ther contention that an autonomous robot capable of lethal force can ultimately be more humane in the battlefield than human soldiers. Robot architectural design recommenda- tions are presented for (1) post facto suppression of unethical behavior, (2) behavioral design that incorporates ethical constraints from the onset, (3) the use of affective func- tions as an adaptive component in the event of unethical action, and (4) a mechanism in support of identifying and advising operators regarding their ultimate responsibility for the deployment of such a system. This research was supported under a grant from the Army Research Office.

25 Tuesday, December 11, 2012

Tu1-1 Humanoid Robots I Tu1-2 Legged Robots I Tu1-3 Vision/Image Processing I Tu1-4 Biologically Inspired Robotics I Tu1-5 Grasping and Manipulation I Tu1-6 Smart Structures, Materials, Actuators I Tu1-7 Service Robots I Tu2-1 Humanoid Robots II Tu2-2 Legged Robots II Tu2-3 Vision/Image Processing II Tu2-4 Biologically Inspired Robotics II Tu2-5 Grasping and Manipulation II Tu2-6 Machine Learning I Tu2-7 SLAM I Tu3-1 Underwater Robots I Tu3-2 Biomimicking I Tu3-3 Medical & Surgery Robots I Tu3-4 Mobile Robots I Tu3-5 Grasping and Manipulation III Tu3-6 Machine Learning II Tu3-7 SLAM II IEEE ROBIO 2012 Conference Digest

Tu1-1: Humanoid Robots I Session Chairs: Haider Sajjad and Yiwei Liu

Room: Jade 1, 10:20-12:05, Tuesday, December 11, 2012

Tu1-1(1) 10:20-10:35 Tu1-1(2) 10:35-10:50

A hierarchical fall avoidance strategy for small- On Evolving a Dynamic Bipedal Walk using scale humanoid robots Partial Fourier Series

Bassam Jalgha, Daniel Asmar, Elie Shammas, and Imad ElHajj Sajjad Haider Mechanical Engineering Department, American University of Beirut Institute of Business Administration, Karachi, Pakistan Beirut, Lebanon Shaukat R. Abidi and Mary-Anne Williams University of Technology, Sydney, Australia • Humanoid robots could benefit from • The paper presents a Partial Fourier a preemptive fall avoidance scheme. Series (PFS) based bipedal gait in • We propose a hierarchical fall sagittal and transverse planes. avoidance strategy including angle, • The gait is optimized through momentum, and stepping strategies Evolutionary Algorithms (EA). as part of the choices the robot can Humanoid taking a step to avoid falling The combination of the two motions make to avoid falling. • enables an adjustable dynamic walk. • The method is hinged upon decision surfaces that help delineate the • The evolved walk is being used optimum strategy corresponding to the magnitude of each disturbance. within the RoboCup Soccer 3D Simulation competition

Tu1-1(3) 10:50-11:05 Tu1-1(4) 11:05-11:20

Innovation for the Lower Limbs of a Humanoid Development of a Face Robot for Cranial Robot Nerves Examination Training Jing-Shan Zhao, Xiang Liu, Fu-Lei Chu and Zhi-Jing Feng Chunbao Wang, Yohan Noh , Hiroyuki Ishii, Atsuo Takanishi Department of Precision Instruments and Mechanology, Tsinghua University, Takanishi Lab, Waseda University, Japan China • Establish rough architecture model for the lower • Novel platform for neurological limb of a examination training. • Investigate the twists of the chain with reciprocal • Various different symptom screw theory reproductions . • Obtain different architecture schemes for the • Provides the trainee more active lower limbs via combining the maximum linearly training. independent screws • Gives better understanding on • Discuss the different schemes for the lower the knowledge of nervous limbs considering the real requirements of Architecture of human being system. The facial expressions engineering with different symptoms

Tu1-1(5) 11:20-11:35 Tu1-1(6) 11:35-11:50 Music conductor gesture recognition by using HIT-ARM I High Speed Dexterous Robot Arm IMU for human-robot musical interaction S. Cosentino1, Y. Sugita1, M. Zecca2, S. Sessa2, Z. Lin3, 3 4 5 Yiwei Liu, Zongwu Xie, Qi Zhang and Bin Wang K. Petersen , H. Ishii , and A. Takanishi 1 State Key Laboratory of Robotics and System School of Advanced Science and Engineering, Waseda University, Japan 2 Harbin Institute of Technology, China School of Creative Science and Engineering, Waseda University, Japan 3 Global Robot Academia, Waseda University, Japan 4 Research Institute for Science and Engineering, Waseda University, Japan 5 Department of Modern Mechanical Engineering, Waseda University, Japan • A dexterous, light weight arm with high • Musical interaction in an orchestra: analysis of precision, high speed manoeuvrability for conductors movements in relation to conveyed playing ping-pong. musical information • Design of modular joint with multisensory • Capture of orchestra conductor hand movements system. using IMU (Inertial Measurement Unit) WB-4 • Hardware architecture of the arm. • Movement data analysis and processing, development of a gesture recognition algorithm Orchestra conductor and Waseda • Experimental results show the reliability of the flutist robot musical interaction proposed gesture recognition system based on IMU

27 IEEE ROBIO 2012 Conference Digest

Tu1-2: Legged Robots I Session Chairs: Zhenli Lu and Mingguo Zhao

Room: Jade 2, 10:20-12:05, Tuesday, December 11, 2012

Tu1-2(1) 10:20-10:35 Tu1-2(2) 10:35-10:50

Simulated performance assessment of different digital A COP-based Controller for Adaptive Motion Planning of hydraulic configurations for use on the HyQ Leg a Single-

Zhenli Lu1,2, Filipe Silva1,3 Yun Zhang 4, Qinruo Wang 4, Luis Seabra Lopes1,3 Shuang Peng, David Branson, Emanuele Guglielmino, and Darwin 1 IEETA, University of Aveiro, Portugal G. Caldwell 2 School of Mechanical Engineering, Shenyang Ligong University, P.R.China Department of Advanced Robotics, Istituto Italiano di Tecnologia, Italy 3 DETI, University of Aveiro, Portugal 4 Guangdong University of Technology • Digital hydraulic systems are able to achieve accurate and smooth responses for specified • Inspired by the motion planning mechanism of motions. human walking gait. • Configurations A and B are able to create • leg robot model with a COP based controller regenerative flow that has an important for velocity planning is developed in YARP. influence on system efficiency, while • Using force sensor to perceive real COP and configurations C and D have better position the changing of environment. tracking. • Robot adopts its motion to changing ground • More valves in each DFCU would improve the Energy consumption of four while realizing some trajectory follow task. Trajectory follow experiment position performance of the system with only a configurations on a rotating platform slight reduction in efficiency.

Tu1-2(3) 10:50-11:05 Tu1-2(4) 11:05-11:20

Analysis of the Disturbance Rejection Ability of Self-Adjusting Locomotion on a Partially a Passive Dynamic Walker with Hip Spring Broken-down Quadrupedal Biomorphic Robot by Evolutionary Algorithms Mingguo Zhao and Biao Hu Guo-Yuan Qiu and Shih-Hung Wu The Department of Automation, Tsinghua University, China Computer Science and Information Engineering, Chaoyang University of Technology, Taichung, Taiwan • The hip spring can enlarge the passive dynamic walker’s basin of attraction. • Dealing with the partially broken-down is an • The hip spring can also increase the important issue for biomorphic robots. A figure here disturbance rejection while keeping existence • We use a PSO-like algorithms to solve this of stable equilibrium state unchanged. problem on . • Arc foot can increase the walking disturbance • We expect that the biomorphic robot can evolve rejection ability greatly only when the foot more self-adjusting behaviors to deal with radius is large. various problems. • Mass distribution has little effect on improving Partially Broken-down Quadrupedal disturbance rejection ability. Passive dynamic walking models Biomorphic Robot

Tu1-2(5) 11:20-11:35 Tu1-2(6) 11:35-11:50

Energetics of Level Walking Trot Gaits in The Function of the Spine and its Morphological Quadruped Robots Effect in Quadruped Robot Locomotion P. Murali Krishna and R. Prasanth Kumar Qian Zhao, Hidenobu Sumioka, Xiaoxiang Yu, Kohei Nakajima, Department of Mechanical Engineering, IIT Hyderabad, INDIA Zhimin Wang, and Rolf Pfeifer Suresh Srivastava GTRE, DRDO, INDIA AI lab, University of Zurich, Switzerland

• This paper investigates energetics of level To test whether or not spine is a dominant factor in locomotion, we walking trot gaits in quadruped robots with A figure here studied the effect of spinal joints on locomotion in terms of attack angle, asymmetric body-mass distribution. CoM, ground clearnce. • Two cases are considered: constant speed, • Advantages of two-spinal joints over one-spinal joint and deceleration-acceleration • Adjust CoM to increase stability • Specific resistance is found to be independent • Offer more freedom to increase stride length of mass distribution • . Advantages of best individual with two spinal joints • Quasi-passive walking is also discussed for Quadruped robot model with • Double flight phases per cycle asymmetric mass-distribution reduced energy consumption • Greater flight proportion

28 IEEE ROBIO 2012 Conference Digest

Tu1-3: Vision/Image Processing I Session Chairs: Junqiu Wang and Xianmin Zhang

Room: Jade 3, 10:20-12:05, Tuesday, December 11, 2012

Tu1-3(1) 10:20-10:35 Tu1-3(2) 10:35-10:50

Unsupervised Progressive Structure-from-Motion for Discriminative Sparse Representation for Online Unordered Images Visual Object Tracking Yingkui Du, Yandong Tang and Jianda Han Shenyang Institute of Automation, Chinese Academy of Sciences, China Tianxiang Bai, Y.F. Li, and Xiaolong Zhou Baojie Fan Dept. of MBE, City University of Hong Kong, Hong Kong University of Posts and Telecommunications, China

• Unordered images are divided into two classes: • We formulate the tracking problem as a binary key frames and non-key frames. classification task. • In key frames, superior features are selected and • A linear classifier is integrated with the sparse tracked to initialize the parameters of PSFM representation framework to form a discriminative reliably and reduce the accumulative errors. model. • A local on-demand scheme is proposed to dramatically reduce the computing cost of sparse • The dictionary and classifier are jointly trained bundle adjustment. using the online dictionary learning algorithm. • Reducing accumulative error efficiently during • The experimental results show that the proposed Representative Tracking scene dense recovery and camera motion method achieves more accurate performance Results estimation Experimental Results than the generative tracker.

Tu1-3(3) 10:50-11:05 Tu1-3(4) 11:05-11:20

Tightly-Coupled Vision-Aided Inertial Navigation Normal Estimation for Pointcloud using GPU via Trifocal Constraints based Sparse Tensor Voting

Ehsan Asadi and Carlo Luigi Bottasso Ming Liu, Francois Pomerleau, Francis Colas, Roland Siegwart Aerospace Engineering Department, Politecnico di Milano, Italy Autonomous Systems Lab, ETH Zurich, Switzerland

Surface normal estimation using sparse • A Tightly-Coupled Vision-Aided Inertial Navigation System (TC-VA-INS) is proposed in this work as a A figure here tensor voting. We compare two synergistic incorporation of vision with other different algorithms in terms of sensors. precision and complexity. We show that • A vision-based model is derived using geometric the proposed algorithm is flexible in constraints among three scenes. structure extraction. A GPU • A kinematic motion model is used to account for implementation of sparse tensor voting. the dynamic behavior of vehicle. We show significant improvement in • Sigma Point Kalman Filtering (SPKF) is used to Geometry for the derivation of the performance comparing with CPU achieve robust estimates in the presence of discrete vision-based model via implementation. nonlinearities. trifocal constraints

Tu1-3(5) 11:20-11:35 Tu1-3(6) 11:35-11:50

Robust Location Recognition Pre-attentive Visual Segmentation Algorithm for Based on Efficient Feature Integration Cognitive Robots

1,3 2,3 Junqiu Wang and Yasushi Yagi Yuanlong Yu and Jason Gu 1 The Institute of Scientific and Industrial Research, School of Automation, Beijing Institute of Technology, Beijing, China 2Shandong University, Jinan, Shandong, China Osaka University, Japan 3Department of Electrical and Computer Engineering, Dalhousie University, Canada

• Multiple feature cues for robust location recognition • Pre-attentive segmentation is an unsupervised • An integrated feature detection: edges and interest points are computed process prior to attentional selection. simultaneously • Bhattacharyya distance based irregular • Characterize different cues using appropriate feature descriptors pyramid (BIP) algorithm • The integration of multi-cue brings efficient feature detection and high • Bhattacharyya distance is used to measure the recognition ratio to the localization system. inter-node similarity • Adaptive neighbor search for each graph at higher pyramidal levels • Improved accuracy and robustness to noise Framework of BIP Algorithm

29 IEEE ROBIO 2012 Conference Digest

Tu1-4: Biologically Inspired Robotics I Session Chairs: Kazuhiro Kosuge and Yisheng Guan

Room: Jade 8, 10:20-12:05, Tuesday, December 11, 2012

Tu1-4(1) 10:20-10:35 Tu1-4(2) 10:35-10:50

A Hybrid Dynamic Model Of An Insect-Like MAV Shape Modeling of a Concentric-tube Continuum Robot With Soft Wings Shaoping Bai and Chuhao Xing Department of Mechanical and Manufacturing Engineering & Centre for Robotics Research Ayman Belkhiri, Mathieu Porez and Frédéric Boyer Aalborg University, Aalborg, Denmark Automation and Production Department, Ecole des Mines de Nantes, France

• Modeling a MAV with compliant wings using • A continuum robot is proposed for robotic surgery inverse hybrid dynamic model. application. • The shape modeling is studied with focus on the • Developing a computationally efficient algorithm torsional deformation on the basis of energy approach. (with highly realistic visual rendering). Both in-plane and out-plane cases are considered. • Computing the net motion of the MAV knowing • The model was experimentally validated. the internal strain fields. • The simulation and experiment results reveal that the • Giving the internal torques required to impose torsional deformation cannot be ignored if the twist of the distal end is concerned. these strain fields along the wing span. A continuum robot and its twist angle at the distal end

Tu1-4(3) 10:50-11:05 Tu1-4(4) 11:05-11:20 Landmark-Tree Map: a Biologically Inspired A lower-limb exoskeleton for gait assistance in Topological Map for Long-Distance Robot quadriplegia Navigation Marcus Augustine1,2 and Elmar Mair2,3 and Annett Stelzer 3 Daniel Sanz-Merodio, Manuel Cestari, Juan Carlos Arevalo and 1 2 3 Elena Garcia and Frank Ortmeier and Darius Burschka and Michael Suppa 1 Otto-von-Guericke University Magdeburg, Germany Field and Service Robotics Group, Centre for Automation and 2 Technische Universität München (TUM), Germany Robotics (Spanish Research Council), Spain 3 German Aerospace Center (DLR), Germany • This paper presents the development and main • a topological map for global navigation inspired features of a lower limb exoskeleton, including: by insect navigation • Synergic biarticular actuation in the ankle. • positions are memorized as unique landmark configurations (snapshots) without metric distance information •Compliance controller based on the force measured by insoles at the feet. • landmarks are organised in a tree (LT-Map), sorted from global to local information • the LT-Map is easily scalable by tree pruning •Parameterized hip and foot trajectories. Simulation of robot movement preserving the dominant information based on Landmark-Tree Map

Tu1-4(5) 11:20-11:35 Tu1-4(6) 11:35-11:50

Viscoelastic Analysis on a Novel Adhesive Tread-Ba Development of MorphHand: Design of an -sed Footpad for Three-Dimensional Climbing Robots Underactuated Anthropomorphic Rubber Finger for a Prosthetic Hand Using Compliant Joints Xuan Wu, Yongjie Zhang, Yanwei Liu, and Chongyang Hu 1,2 2 3 University of Science and Technology of China, PRC Tapio V. J. Tarvainen , Wenwei Yu , and Jose Gonzalez 1 Shaoming Sun and Tao Mei Dept. of Electronics, Aalto University, Finland 2Dept. of Medical System Eng., Chiba University, Japan Hefei Institute of Physical Science, Chinese Academy of Sciences, PRC 3Research Center for Frontier Medical Eng., Chiba University, Japan

• The tread-based footpad for three-dimensional • First steps in the development of MorphHand climbing robots is designed. A figure here and a new manufacturing method for • The self-preloading mechanism of the footpad prosthetics is studied for maximum safety. • Experiments on five 3D-printed two-joint rubber • Detachment of the footpad is studied to reveal fingers with nine different configurations the viscoelastic mechanical property. • Acquired essential information on the fingers’ • Experiments verify the analysis and the morphological parameters and performance for The self-preloading of the footpad Prototype rubber finger in CAD footpad prove to be promising for climbing further development robots. and during an experiment

30 IEEE ROBIO 2012 Conference Digest

Tu1-5: Grasping and Manipulation I Session Chairs: Yong Yu and Dan Zhang

Room: Jade 5, 10:20-12:05, Tuesday, December 11, 2012

Tu1-5(1) 10:20-10:35 Tu1-5(2) 10:35-10:50

Design and Analysis of a Miniature 4- Dynamic Analysis of Planar 3-RRR Flexible Dimensional Force/Torque Sensor parallel robots Qiaokang Liang Qinghua Zhang, Xianmin Zhang, Jinglun Liang College of Electrical and Information Engineering, Guangdong Province Key Laboratory Precision Equipment and Manufacturing Hunan University, Changsha, Hunan, 410082, China Technology, South China University of Technology Dan Zhang • Based on the complex mode me- QingDao Technological University, Qingdao China thod, the dynamic characteristics • The design of a 4-dimensional fingertip F/T of the systems that include the fr- sensor is proposed equency characteristics, the sen- • The Elastic Element (EE) structure featured sitivity, modes and dynamic stre- with an annular diagram and flexible beams, sses are studied though numeric- and its static and dynamic performances are al simulation. analyzed. • Time-varying nonlinear second- • The proposed sensor has the advantages of order differential equation is Planar 3-RRR flexible parallel robot The gripper equipped with the developed simple and miniature configuration, high proposed 4-D fingertip F/T sensors sensitivity, and isotropy.

Tu1-5(3) 10:50-11:05 Tu1-5(4) 11:05-11:20

TACTIP - Tactile Fingertip Device, Challenges in Biologically inspired tactile classification reduction of size to ready for robot hand of object-hand and object-world interactions Benjamin Winstone and Gareth Griffiths Barrett Heyneman and Mark R. Cutkosky Bristol Robotics Laboratory, University of the West of England, UK Mechanical Engineering, Stanford University, USA Chris Melhuish Bristol Robotics Laboratory, University of the West of England, UK • Designed a suite of capacitive tactile • Biologically-inspired sensing device, based sensors with bandwidth of 150Hz. upon the deformation of the epidermal layers of • Developed a classification method based the human skin. on relative signal power between individual Object-hand • Object interaction measured optically by taxels and ensembles. tracking internal papillae pins. • Achieved 84% successful classification of • Papillae pins are representative of the slip location for a range of textures and

intermediate epidermal ridges of the human situations. Object-world skin. Cross section of TACTIP device • Demonstrated insensitivity to the frequency Gripper with tactile array sensor band in which the power ratio is computed. discriminating between slip locations

Tu1-5(5) 11:20-11:35 Tu1-5(6) 11:35-11:50 Optimization for Lower Limb Assistive Mechanism Based on Manipulability Inclusive Gripper Self-Alignment for Autonomous Principle Pole-Grasping with a Biped Climbing Robot 1 2,3 Yong Yu and Wenyuan Liang Yisheng Guan, Zhiguang Xiao, Wenqiang Wu, Junjun Wu, Haifei Zhu and Manjia Su 1. Graduate School of Science & Engineering, Kagoshima University, Japan Biomimetic and Intelligent Robotics Lab (BIRL) 2. Institute of Intelligent Machines, Chinese Academy of Science, China School of Mechanical and Automotive Engineering 3. Dept. of Automation, University of Science and Technology of China South China University of Technology

• Design optimization for lower limb  A sensing system consisting of multiple assistive mechanism is based on sensors to detect the relative configuration Manipulability Inclusive Principle (MIP); of the gripper is developed • By comparing the manipulability of  An algorithm based on the sensed data to assisted limb and slave-active-assistive adjust the gripper configuration for align- mechanism, MIP can evaluate assistive ment with the target pole is proposed feasibility and assistive effect;  The integrated gripper can autonomously • Realizing assistive feasibility & better grasp a fixed pole and follow a moving pole assistive effect can be obtained through  The self-alignment method may be used for the optimization algorithm based on autonomous climbing with climbing robots MIP. and grasping with industrial robots Manipulability Inclusive Principle

31 IEEE ROBIO 2012 Conference Digest

Tu1-6: Smart Structures, Materials, Actuators I Session Chairs: Huayan Pu and Shinichi Hirai

Room: Jade 9, 10:20-12:05, Tuesday, December 11, 2012

Tu1-6(1) 10:20-10:35 Tu1-6(2) 10:35-10:50 Experimental Study on Oscillating Paddling Achieving Compliant Spherical Linkage Designs Gait of an Eccentric Paddle Mechanism from Compliant Planar Linkages Based on Huayan Pu, Yi Sun, Shugen Ma, Zhenbang Gong PRBM: A Spherical Young Mechanism Case •Eccentric Paddle Mechanism Study • a moveable paddle shaft Li Geng and Chen Guimin • a set of four paddles School of Mechatronics, Xidian University, China • an actively actuated shell • four passive paddle hinges • A planar linkage can be considered as a special •Oscillating Paddling Gait spherical linkage (with an infinite spherical A figure here radius). • The rigid-body replacement synthesis approach can be employed to achieve compliant designs •Experimental Study On Oscillating Gait of the PRBM. • A Young mechanism is changed into a spherical • Oscillating Paddling gait can generate the effective underwater thrust • Amplitude and period time of oscillation trajectory affect the characteristics of fluid force Young mechanism, and is proved has bistability Spherical Young mechanism in both time domain and frequency domain behaviors. • Non-working paddles play a negative role on the generation of net thrust

Tu1-6(3) 10:50-11:05 Tu1-6(4) 11:05-11:20

Dynamic Simulation of An Asteroid Landing Mechanism and Six-strut Tensegrity Robot Rolling Its Landing Simulation Zhijun Zhao, Jingdong Zhao and Hong Liu Shinichi Hirai and Ryo Imuta State Key Laboratory of Robotics and System, Department of Robotics, Ritsumeikan University, Japan Harbin Institute of Technology, China

• Rolling of six-strut tensegrity robot. • Introduction. • Dynamic simulation of rolling. • Mechanical structure of the asteroid landing • Constraint stabilization method formulates mechanism. contact between struts and ground. • Theoretical analysis of the landing dynamic. • Comparison between simulation and • Landing Simulation with different constraints. experimental results in tensegrity robot rolling. • Conclusion.

Model of six-strut tensegrity robot Simulation model and its results

Tu1-6(5) 11:20-11:35 Tu1-6(6) 11:35-11:50

A valveless piezoelectric pump with multistage Design of High Capacity Electro Permanent Y-shape treelike bifurcate tubes Magnetic Adhesion for Climbing Robots

Jun Huang and Jianhui Zhang Peter Ward and Dikai Liu State Key Laboratory of Mechanics and Control of Mechanical Structures, Faculty of Engineering, University of Technology, Sydney, Australia Nanjing University of Aeronautics and Astronautics, China

• A valveless piezoelectric pump with OFF multistage Y-shape treelike bifurcate • High capacity electro permanent magnets are tubes was designed. designed for use in climbing robots. • The theoretical analysis of flow resistance • Theoretical analysis, simulations and characteristics and flow rate were performed. experiments demonstrate their feasibility. • This valveless piezoelectric pump was • Benefits include being power-fail safe, fast fabricated, and experiment of flow rates was switching times, low power and light weight. investigated. valveless piezoelectric pump with multistage Y-shape treelike ON bifurcate tubes

32 IEEE ROBIO 2012 Conference Digest

Tu1-7: Service Robots I Session Chairs: Genci Capi and Jianwei Zhang

Room: Jade 7, 10:20-12:05, Tuesday, December 11, 2012

Tu1-7(1) 10:20-10:35 Tu1-7(2) 10:35-10:50 Redundant Actuation Research of the Development of a New Robotic System for Quadruped Walking Chair with Parallel Leg Assisting and Guiding Visually Impaired People Mechanism Lingfeng Sang, Hongbo Wang, Xiong Zhang, Xiangwang Kong G. Capi and Yan Liang Dept. of Electric and Electronic Eng., University of Toyama, Toyama, Japan Parallel Robot and Mechatronic System Laboratory , Yanshan University, • This paper presents a new intelligent robotic China system to assist or guide visually impaired Dianfan Zhang people in unknown indoor environments. Science and Technology Administration Office , Yanshan University, China • The robotic system, which is equipped with a visual sensor, laser range finders, speaker, Developed system. • Design idea and coordinate system setting of gives visually impaired people information walking chair about the environment around them. • Input selecting theory • In addition, the robot can be switched to the guide mode. By selecting the target location, • Constraint screw of the branch and the the robot starts guiding the user toward the actuation input combination goal utilizing the line on the ground or • Actuating input evaluation and some evolved neural controllers. The usefulness of the proposed system is examined conclusions The quadruped walking chair experimentally. Video capture of the experiment.

Tu1-7(3) 10:50-11:05 Tu1-7(4) 11:05-11:20

Racket Control and its Experiments for Robot New Approach in Control of Playing Table Tennis Assembly/Disassembly Line Served by Robotic Mounted on Mobile Platform Chunfang Liu, Yoshikazu Hayakawa and Akira Nakashima E. Minca, A. Filipescu and A. Voda Department of Mechanical Science and Engineering, Nagoya University, Japan Department of Automation and Electrical Enginering University “Dunarea de Jos” of Galati,Romania •The new idea of this paper is to make • Propose a method for table tennis robot to reversible an assembly line, i.e. to allow complete disassembly, by using a mobile control the racket’s state at the striking point. platform equipped with robotic manipulator. • It can return a spinning ball to a destination •The approach is a hybrid one in which the with desired rotational velocities at a desired High Speed Cameras assembly/disassembly line is the discrete (900fps) system whilst the wheeled 7-DOF Manipulator landing time. Automatic Ball (WMR) together with the robotic manipulator Catapult (RM) is considered the continuous one. • Derived by solving racket rebound model

•The mobile platform is used only in z and two point boundary value problem of y disassembling operations, in order to transport x the components from the disassembling aerodynamics model. locations to the storage locations. Therefore, Synchronized Hybrid Petri Nets approach is • Propose simple aerodynamics model. A robot is playing table tennis Control structure of assembly/disassembly line used in modeling and control. served by WMR equipped with RM Experiments verify the effectiveness of the proposed algorithms.

Tu1-7(5) 11:20-11:35 Tu1-7(6) 11:35-11:50

Stress Relief Robotic System Based on Diffused Design of Robot for Space Station Operation Illumination Multi-touch Technology Xin Tian 1 1 1,2 1,2 Department of Computer Science,University of Science and Technology of Yongquan Chen , Yong Yang , Huihuan Qian ,Yangsheng Xu China,China Dept. of Mechanical and Automation Engineering, CUHK, Hong Kong1 Institutes of Advanced Technology, CAS, Shenzhen, China2 Xinyu Wu, Jianquan Sun, Ping Jiang, Ruiqing Fu Center for Intelligent and Biomimetic Systems, ShenZhen Institutes of Advanced Technology,Chinese Academy of Sciences, China • Motivation of new robot for space station • Respond to user input such as voice and • Rail design touch. • Mobile base design • Give out graphical and speech respondence. • Robot arm design • The system is scalable. New applications can • Robot dynamics simulation and FEA analysis be developed and integrated into the system for space operation tasks easily. • A Tomcat-like application is developed as a proof-of-concept demo.

33 IEEE ROBIO 2012 Conference Digest

Tu2-1: Humanoid Robots II Session Chairs: Yili Fu and Jingwei Guan

Room: Jade 1, 13:30-15:30, Tuesday, December 11, 2012

Tu2-1(1) 13:30-13:45 Tu2-1(2) 13:45-14:00

Footprint Searching and Trajectory Design of a Robust Swing Leg Placement Humanoid Robot under Large Disturbances Pin-Yong Ling, Kenneth Yi-wen Chao, Han-Pang Huang, Ruta Desai and Hartmut Geyer The Robotics Institute, Carnegie Mellon University, USA and Jiu-Lou Yan Department of Mechanical Engineering, National Taiwan University, Taiwan • We present a powerful alternate for • A footstep-planning algorithm proposed for robust swing leg placement in autonomous navigation in 3D environments for humanoids and rehabilitation robotics. humanoid robots. A figure here • The control relies on local feedback A figure here and does not require any predefined • Taking variant heights, horizontal coordinates, trajectories. and possible footsteps into account in the The control shows human like footstep-planner. • response to disturbances during swing. • Locomotion execution with re-planning scheme • Further, it generates swing leg for correcting navigation errors. trajectories and joint torques similar to • Modifications of momentum compensation for The humanoid robot rendered in human walking and running at different Observed and predicted (A) ankle irregular footsteps or tuning motions. Adams for one of the simulations speeds (see figure). trajectories (B&C) hip & knee torques in walking, respectively.

Tu2-1(3) 14:00-14:15 Tu2-1(4) 14:15-14:30 FPGA-implementation of Passivity-based Design And Development of A 7-DOF Control for Elastic-joint Robot Humanoid Arm

Qi Zhang, Zongwu Xie, Hegao Cai, and Hong Liu Xijian Huo, Yiwei Liu, Li Jiang and Hong Liu State Key Laboratory of Robotics and System, State Key Laboratory of Robotics and System Harbin Institute of Technology, China Harbin Institute of Technology, China

• Overview of the humanoid robot system • Theory analysis and simulation of passivity- A figure here • A 7-DOF humanoid arm. based control for elastic-joint robot • The evaluation of 7-DOF arm configurations. • Realization of the controller on FPGA • Design of modular joints with multisensory • Experiment and results on a 7-dof robot systems. manipulator. • The communication architecture of 7-DOF humanoid arm. Structure of joint controller in FPGA

Tu2-1(5) 14:30-14:45 Tu2-1(6) 14:45-15:00

Study on Distance Measurement for NAO Designing Gestures with Semantic Meanings for Humanoid Robot Humanoid Robot Jingwei Guan Minhua Zheng and Max Q.-H. Meng School of Control Science and Engineering, Shandong University, China Department of Electronic Engineering, The Chinese University of Hong Kong, Max Q.-H. Meng Hong Kong, China Department of Electronic Engineering, The Chinese University of Hong Kong, China • Semantic gestures are extracted from daily life for human-robot interaction and translated from human to humanoid robot Nao. Four different methods are used to • • People’s perceptions of semantic gestures on measure the distance information. both human and humanoid robot are tested. The methods are sonar, stereo vision, • • Comparative studies indicate people’s Asus Xtion Pro Live and NDI. perception difference in the two cases. The study is aimed at finding the most • • A framework for designing and evaluating appropriate way to measure the depth. semantic gestures on humanoid robot is System Environment proposed.

34 IEEE ROBIO 2012 Conference Digest

Tu2-2: Legged Robots II Session Chairs: Shuro Nakajima and Kenichi Narioka

Room: Jade 2, 13:30-15:30, Tuesday, December 11, 2012

Tu2-2(1) 13:30-13:45 Tu2-2(2) 13:45-14:00

Concept of Adaptive Gait for Leg-wheel Stable Running Velocity Change of Biped Robot Based on Virtual Torque Robot, RT-Mover Jae Uk Cho, Je Sung Yeon, and Jong Hyeon Park Shuro Nakajima and Kazuhisa Ietomi Department of Mechanical Engineering, Department of Advanced Robotics, Chiba Institute of Technology, JAPAN Hanyang University, Seoul 133-791, Korea RT-Mover can move on This paper proposes a method to continuous rough terrain using a • wheel mode. This paper generate a stable running trajectory for a proposes concept of adaptive gait biped robot based on virtual torque to using leg functions. When the change a velocity. wheels come in contact with convex terrain, adaptive gait will • A virtual torque at the supporting foot for allow movement to continue a velocity change is generated by through the use of leg motions. moving the ZMP. This paper provides a proposal • Based on the GCIPM with virtual torque, concerning the basic operation of adaptive gait and validates a trajectory for stable velocity changes is their efficacy using the results of generated. simulations and experiments.

Tu2-2(3) 14:00-14:15 Tu2-2(4) 14:15-14:30

Development of a Minimalistic Pneumatic The Walk-&-Roller: A Multimodal Robotic Vehicle Quadruped Robot for Fast Locomotion Capable of Quasistatic Three-Legged Rolling Motion Kenichi Narioka, Andre Rosendo, Koh Hosoda Mahmood Karimi, Ky Woodard, Benjamin A. Sams, Thomas R. Bewley Coordinated Robotics Laboratory, University of California, San Diego Grad. School of Information Science and Technology, Osaka University, Japan. La Jolla, CA 92093 USA Alexander Sproewitz School of Engineering, Ecole Polytechnique Federale de Lausanne, Switzerland

• A new quasi-static gait for a 3-legged • The quadruped robot ”Ken” were developed robot named the Walk-&-Roller (W&R) with the minimalistic and lightweight body • The walking is achieved through rotation design for achieving fast locomotion. of the main body of the robot • McKibben pneumatic artificial muscles are • The trajectory planning is performed in such a way that the static tipover stability used as actuators, providing high frequency is guaranteed and no singularities are and wide stride motion of limbs, also avoiding reached. problems with overheating. • Experiments have been performed to demonstrate the validity of this proposed • A preliminary experiment were conducted, gait method finding out that the robot can swing its limb over 7.5 Hz without amplitude reduction.

Tu2-2(5) 14:30-14:45 Tu2-2(6) 14:45-15:00

Stepping to Recover: A 3D-LIPM Based Push Recovery A Miniature Biped Wall-Climbing Robot for and Fall Management Scheme for Biped Robots Inspection of Magnetic Metal Surfaces Awais Yasin, Qiang Huang, Zhangguo Yu, Qian Xu and Zhiqiang Bi, Yisheng Guan, Shizhong Chen, Haifei Zhu and Hong Zhang1,2 Amjad Ali Syed 1 Biomimetic and Intelligent Robotics Lab (BIRL) Intelligent Robotics Institute, School of Mechatronics Engineering, Beijing School of Mechanical and Automotive Engineering Institute of Technology, Beijing, P.R. China South China University of Technology, China 2 Dept. of Computing Science, University of Alberta, Canada • 3D Linear Inverted Pendulum Model for Biped Walk.  A miniature biped wall-climbing robot is developed for ultrasonic inspection of magnetic • Modeling the robot for push recovery. metal surfaces • Different push recovery schemes.  The robotic system and the adhesion force's • Push recovery and fall avoidance through analysis are presented stepping.  Three climbing gaits and one detection gait are • Fall management for safe and customized illustrated with a series of experiments falling. Push Recovery and Fall  A potential application with MiniBibot-W is Management demonstrated in ultrasonic detection

35 IEEE ROBIO 2012 Conference Digest

Tu2-3: Vision/Image Processing II Session Chairs: Wei Yang and Ying Liu

Room: Jade 3, 13:30-15:30, Tuesday, December 11, 2012

Tu2-3(1) 13:30-13:45 Tu2-3(2) 13:45-14:00 Histogram of Silhouette Direction Code: An Efficient Probabilistic Block-Matching Based 6D Camera HOG-based Descriptor for Accurate Human Localization Detection Damien Vivet, Benoit Priot, and Vincent Calmettes Wei Yang, Zhan Song and Xinyu Wu ISAE, Universités de Toulouse, France Shenzhen Institute of Advanced Technology, Chinese Academy of Science Clément Deymier Institut Pascal, Clermont Université, France • Presented a novel discrimination function of - Full 6D localization algorithm based on motion the difference in brightness for a better field. background subtraction. - Adaptation of MPEG Block-Matching for sparse • Proposed HSD feature which uses the optical flow. histogram of direction code along the silhouette as descriptor. - Probabilistic analyses of motion flow and extraction of uncertainty. • Experimental results with real video sequences show its improvements in both segmentation - Uncertainty propagation for 6D probabilistic efficiency and accuracy. localization.

Tu2-3(3) 14:00-14:15 Tu2-3(4) 14:15-14:30

Moving Target Detection based on the Optimal Quantization of JPEG2000 Baseline Properties of Corners Compression in RGB Space Ying Liu, Yimin Zhou and Guoqing Xu Bin Leng, Qing He, Dewen Zeng Shenzhen Institutes of Advanced Technology, Intelligent control research center, Guangzhou Institutes of Advanced Chinese Academy of Sciences, Technology, Chinese Academy of Sciences, China Shenzhen, China, 518055

Proposed Inter-frame Regional Corners • • A novel mathematical model for Difference to extract the object area of quantization ratio among the three A figure here effective movement. It can effectively color component in RGB is developed. save computing time and eliminate the VDR (voyage data recorder) test static background; • methods combined with image quality • The velocity constraint is put forwarded are used to evaluate the optimal to extract the moving target, the compression ratio, considering human constraint can eliminate the dynamic eye characteristics. background effectively. Experimental results comparison JPEG2000 pipeline scheme for image compression

Tu2-3(5) 14:30-14:45 Tu2-3(6) 14:45-15:00

Robust Optical flow Estimation for Illumination Monocular Vision Tracking Based On Particle Changes Filter and Hu Moment Xiuzhi Li , Xue Zhao, Songmin Jia Xiuzhi Li, Songmin Jia and Xue Zhao College of Electronic Information & Control Engineering, College of Electronic Information & Control Engineering, Beijing University of Beijing University of Technology, Technology, CHINA Beijing, China • How to tackle the difficulty involved in optical flow calculation caused by illumination • This paper presents a monocular vision inconsistency in adjacent video frames remains tracking approach based on Hu moment and an open challenge. Particle Filter. • This paper investigates how to obtain a valid • Hu moment is used as identification features and accurate optical flow field which is robust for its inherent invariance property. to illumination change based on structure- • PF is adopted to correct and predict the texture decomposition. location of the tracking object. • The advent of this idea for optical flow calculation is verified by experiment results.

36 IEEE ROBIO 2012 Conference Digest

Tu2-4: Biologically Inspired Robotics II Session Chairs: Tetsuro Funato and Shinya Aoi

Room: Jade 8, 13:30-15:30, Tuesday, December 11, 2012

Tu2-4(1) 13:30-13:45 Tu2-4(2) 13:45-14:00 Multiscale Dynamic Modeling of Flexibility in Design of An Anthropomorphic Tendon-Driven Myosin V using a Planar Mechanical Model Robotic Finger Mahdi Haghshenas-Jaryani, and Alan Bowling Shouhei Shirafuji, Shuhei Ikemoto and Koh Hosoda The Robotics, Biomechanics, and Dynamic Systems Lab., UTA, Arlington, USA Multimedia Engineering, Graduate School of • A new multiscale dynamic model of Information Science and Technology, Osaka University, Japan nanoscale Myosin V with flexible necks. • Design of a tendon-driven mechanism of an • The proposed model addresses the anthropomorphic robot hand. extra disproportionate terms in the • We use an equivalent model in the joint torque EOM that cannot be handled by the caused by muscle tensions to reconstruct the original approach. function of human fingers. • The proposed model retains mass. • Proposed model makes it easy to apply the The results present the consistency structure of the human finger to a tendon- • A Schematic of the modified model with the driven robotic hand with actuators fixed to the Representation of original one in prediction of the base and a tendon- pulley system. Myosin V physical characteristics.

Tu2-4(3) 14:00-14:15 Tu2-4(4) 14:15-14:30

A System Model That Focuses on Kinematic Synergy for Cusp catastrophe embedded in gait transition of Understanding Human Control Structure a quadruped robot driven by Tetsuro Funato, Nozomi Tomita and Kazuo Tsuchiya nonlinear oscillators with phase resetting Dept. of Energy and Mechanical Engineering, Doshisha University, Japan Shinya Aoi1, Daiki Katayama1, Soichiro Fujiki1, Takehisa Kohda1, Shinya Aoi Kei Senda1, and Kazuo Tsuchiya2 Dept. of Aeronautics and Astronautics, Kyoto University, Japan 1 Kyoto University, Japan, 2 Doshisha University, Japan

• Walking control model including dimension- • Quadruped robot whose legs are controlled reduction is proposed as a possible human by nonlinear oscillators with phase resetting. strategy of locomotion control. • Walk–trot transition with a hysteresis occurs • Low-dimensional structure of kinematics during by changing the locomotion speed. walking (kinematic synergy) is obtained from • Dependence of the gait transition on the human experimental data. physical conditions, such as body mass and • Simulation of the proposed control model using leg length, is investigated. human kinematic synergy realized walking on • Codimension-2 cusp bifurcation appears in Cusp catastrophe in gait transition flat and slope floors. the gait transition. for duty factor and body mass

Tu2-4(5) 14:30-14:45 Tu2-4(6) 14:45-15:00

Self-Stabilizing Function of Two Dimensional Improving adaptive walking of Human Lower Limb Musculoskeletal System a biped robot on a splitbelt treadmill by controlling the interlimb coordination Handdeut Chang, Yasuhiro Sugimoto, Koichi Osuka 1 1 1 2 Mechanical Engineering, Osaka University, Japan Soichiro Fujiki , Shinya Aoi , Kei Senda , and Kazuo Tsuchiya Yoshiyuki Sankai 1Kyoto University, Japan 2Doshisha University, Japan Information Engineering, Tsukuba University, Japan • We conducted computer simulation • We investigate how a musculoskeletal system is of splitbelt treadmill walking of a stabilized by its own properties with no control. biped robot. • Our previous control system using nonlinear oscillators with phase • Negative gradient of force-length, force-velocity resetting produced adaptive walking. curve of muscle is important for self-stability of static musculoskeletal system. • We introduced the control of the interlimb coordination into the control • Importance of elasticity of muscle in dynamic Two dimensional musculoskeletal system. musculoskeletal system is analytically shown by model and Hill-type muscle model • It improved adaptive splitbelt Relative phase between the legs Lyapunov stable theory. treadmill walking of the robot.

37 IEEE ROBIO 2012 Conference Digest

Tu2-5: Grasping and Manipulation II Session Chairs: Johannes Alexander Coy and Dikai Liu

Room: Jade 5, 13:30-15:30, Tuesday, December 11, 2012

Tu2-5(1) 13:30-13:45 Tu2-5(2) 13:45-14:00

Motion Control of Instantaneous Force for an Artificial Muscle Manipulator with Variable Rheological Joint Development of a Robot for Transferring Mycobacterium tuberculosis Samples Hiroki Tomori, Suguru Nagai and Tatsuo Majima Department of Precision Mechanics, Chuo University, Japan Yan Wang , Guanyang Liu and Chao Chen Taro Nakamura Robotics Institute, Beihang University, China Chuo University, Japan

• We developed a 1-DOF manipulator using a pneumatic artificial muscle and a MR brake. • Automatic Tuberculosis (TB) testing/diagnosis • We proposed a method for generating system. instantaneous force by the artificial muscle. • 3-DOF robot arm for transferring TB samples. • We controlled the manipulator arm motion with • Closed loop control Method of stepping instantaneous force by controlling the MR motors. brake. • Software interface for programming the work 1-DOF artificial muscle steps. manipulator

Tu2-5(3) 14:00-14:15 Tu2-5(4) 14:15-14:30

A Comparative Study of H∞ and PID Control for Analysis of a contact problem for tactile sensors and Indirect Deformable Object Manipulation a computationally effective simulation for grasping Dmitry Sinyukov, Taskin Padir, and Eduardo Torres-Jara Steven Kinio and Alexandru Patriciu Robotics Engineering, Worcester Polytechnic Institute, USA Electrical and Computer Engineering, McMaster University, Canada

• Compares the performance of a standard PID • Object: elastic hemi-sphere tactile sensor for controller and a robust H∞ controllers large sensor arrays • Controllers are designed to perform the task of A figure here • Simplification 1: Linear first order sensor indirect deformable object manipulation abstraction (zero-mass rigid body on a dumped • Controllers are generated using finite element spring) model of deformable object • Simplification 2: Virtual segment method for • H∞ controller performed superiorly compared contact simulation => linear contact model to tuned PID in experimental setting • Result: Increase in simulation performance Sensor, sensor model, compared to a finite-element model by several contact model, simulation orders.

Tu2-5(5) 14:30-14:45 Tu2-5(6) 14:45-15:00

Studies of Adaptive Control Methods Based on VSC First Step Towards a Flexible Design Kit for Trajectory Tracking of Robotic Manipulators Johannes A. Coy, Jan D. J. Gumprecht, Jingmei Zhai, Haiyang He and Bo Kang College of Mechanical Engineering, South China University of Technology, Daniel B. Roppenecker, and Tim C. Lüth China Faculty of Mechanical Engineering, Technical University of Munich, München, Germany

• A PID Plus Compensation Based • Flexible Design Kit for laparoscopic grippers on VSC is proposed. A figure here • Optimized for selective laser sintering • An Adaptive Fast Control Method Based on • Based on compliant mechanism (a) VSC is presented. • First experiments to determine the behavior of • The effectiveness of both methods has been small sintered compliant mechanisms confirmed by simulations and experiments. • Models predicting the behavior could not (b) • Comparisons has been made between current be proved control methods and the proposed methods. Trajectory of the end of the Concept of the Design Kit (a) robotic manipulator Deformed Element (b)

38 IEEE ROBIO 2012 Conference Digest

Tu2-6: Machine Learning I Session Chairs: Han Wang and Yiannis Gatsoulis

Room: Jade 9, 13:30-15:30, Tuesday, December 11, 2012

Tu2-6(1) 13:30-13:45 Tu2-6(2) 13:45-14:00

7-Motion Discrimination Technique for Forearms Classifier ensemble with incremental learning Using Real-Time EMG Signals for disaster victim detection

Nobutaka Tsujiuchi , Haruaki Mizuno, Takayuki Koizumi and Bhuman Soni and Arcot Sowmya School of Computer Science and Engrineering, Shinya Yamada University of New South Wales Department of Mechanical Engineering, Doshisha University, Japan Australia

Motion2 Decision Region • We developed a method of discriminating Motion2 • Human Victim detection using classifier ensembles with HOG and SURF features. real-time motion from electromyogram (EMG) Motion1 Decision Region A figure here signals. • Use of detection history as well as majority • We proposed Hyper-Sphere model. voting to resolve classification outcome • This model can discriminate seven hand • Incremental online training of the classifiers in

motions in real-time. Motion1 the ensemble. • Conduct online training without interrupting the • Elbow motions don’t interfere with the hand ・・・Center of the sphere motion discrimination. ongoing classification process. Incremental learning Hyper-Sphere Model process

Tu2-6(3) 14:00-14:15 Tu2-6(4) 14:15-14:30

Agricultural Produce Grading by Computer Closed-loop Multiple View Registration Vision Using Genetic Programming Xiaowei Shao, Yun Shi, Yulin Duan and Ryosuke Shibasaki Institute of Industrial Science, University of Tokyo, Japan Panitnat Yimyam and Adrian F. Clark Huijing Zhao Computer Science and Electronic Engineering, University of Essex, UK State Key Lab of Machine Perception, , China • We present an approach to generating task- • Address multiple view registration under a loop specific computer vision systems from generic constraint via iterative update on manifold. components using machine learning • Precisely decouple the rotation and translation • The approach generates solutions to first the components of multiple poses. segmentation of features and then their • Provide a theoretical view for understanding the classification mechanism of update propagation in a loop. • Experiments are performed on grading several • Provide an accurate, stable and efficient Convergence and accuracy different types of produce, and the results numerical solution, even for 1000 views. when handling 1000 views compared with what can be achieved using Sample agricultural produce other classification approaches with different quality

Tu2-6(5) 14:30-14:45 Tu2-6(6) 14:45-15:00

Biologically Inspired Intrinsically Motivated Learning for A Fast and Robust Head Pose Estimation Service Robots Based on Novelty Detection and Habituation System Based on Depth Data Y. Gatsoulis*, T.M. McGinnity* and C. Burbridge# *University of Ulster, #University of Birmingham Xiaozheng Mou and Han Wang School of Electrical & Electronic Engineering, Nanyang Technological University, Singapore

 Intrinsic motivation (IM) has received attention • KF is used to predict head location and limit  Use biological mode of habituation as IM the search space of DRRF. A figure here in cumulative learning • The result of DRRF is taken as the  Experiments with physical PR2 learning measurement of KF. perceptions of objects • Head pose of current frame is computed by KF  Results showed the ability of the system and used for next frame’s prediction. to learn autonomously novel objects • The proposed algorithm is faster, more robust Flowchart of the proposed and more accurate than the original DRRF. algorithm

39 IEEE ROBIO 2012 Conference Digest

Tu2-7: SLAM I Session Chairs: Aiguo Ming and Ioannis Rekleitis

Room: Jade 7, 13:30-15:30, Tuesday, December 11, 2012

Tu2-7(1) 13:30-13:45 Tu2-7(2) 13:45-14:00 A Novel Mapping Strategy based on Neocortex Monocular SLAM: Using Trapezoids to Model Model: Pre-liminary Results by Hierarchical Landmark Uncertainties Temporal Memory Xinzheng ZHANG and Jianfen ZHANG School of Electrical and Information Engineering, Jinan University, China M. Hossein Mirabdollah and Bärbel Mertsching Faculty of Electrical Engineering, Computer science and Mathematics, Ahmad B. RAD Paderborn University, Germany School of Engineering Science, Simon Fraser University, Canada

• Propose a mapping method based on the • Given bearing-only measurements, hierarchical temporal memory network A figure here uncertainties of the measurements can be • Treat the mapping problem as the recognition mapped into trapezoids to model the problem uncertainties of landmark positions. • Construct a 2-level HTM network for mapping • Uncertainty of a robot position is modeled by application including loop closure detection. weighted particles, to each of which M • The format of the environmental map is a trapezoids for M landmarks are attached. sequence of scenes which are formed by the For figure caption (optional), • Trapezoids and the robot position are trimmed Trimming principle of combination of visual word projected from SURF. use arial narrow 20pt font simultaneously, using a recursive Bayesian trapezoids filter.

Tu2-7(3) 14:00-14:15 Tu2-7(4) 14:15-14:30

Improving 3D Indoor Mapping Real-time Monocular Visual Self-localization Approach Using with Motion Data Natural Circular Landmarks for Indoor Navigation Jianhao Du and Weihua Sheng Yuandong Sun, Ning Ding, Huihuan Qian, and Yangsheng Xu School of ECE, Oklahoma State University, USA Advanced Robotics Lab., The Chinese University of Hong Kong, Hong Kong, China Yongsheng Ou Shenzhen Institute of Advanced Technology, China • Using natural landmarks on the ceiling. • A 3D mapping method is proposed using RGB-D • Simple but effective. Feasible for sensor data and motion data. indoor environment, especially • The relative motion is estimated by motion data some office environment. based on the offset from the calibration. • Capable for real-time application. • A multiple ICP framework is proposed to improve the robustness. • The results are compared with other existing methods. 3D mapping for an indoor environment Circular Landmarks on the Ceiling Feature selection

Tu2-7(5) 14:30-14:45 Tu2-7(6) 14:45-15:00

Mobile Robot Parallel PF-SLAM Simultaneous Localization and Uncertainty Based on OpenMP Reduction on Maps (SLURM): Ear based Exploration Songmin Jia, Xiaolin Yin and Xiuzhi Li Ioannis Rekleitis College of Electronic Information & Control Engineering, School of Computer Science, McGill University, Canada Beijing University of Technology, Beijing, China

• Single robot explores an environment equipped with a Camera Sensor Network. A figure here • This paper presents an effective Simultaneous • Systematic loop closure, exploring one face at Localization and Map Building technique for a time, reduces the uncertainty buildup. indoor mobile robot. • Exploration versus exploitation strategies used • A multi-thread particles filter algorithm based to decide when the robot would explore new on OpenMP is proposed. areas and when it would reduce the Ear based exploration: The • The results show that the algorithm could robot made only clockwise reduce SLAM execution time. uncertainty in the known map. turns until returning to an explored node.

40 IEEE ROBIO 2012 Conference Digest

Tu3-1: Underwater Robots I Session Chairs: Lige Zhang and Gert Toming

Room: Jade 1, 15:20-17:05, Tuesday, December 11, 2012

Tu3-1(1) 15:20-15:35 Tu3-1(2) 15:35-15:50

Underwater Object Detection and Localization Line formation algorithm in a self-organized Based on Multi-Beam Sonar Image Processing swarm of micro-underwater unmanned vehicles Li-Yuan Weng, Min Li and Zhenbang Gong Thomas Sousselier School of Mechatronic Engineering and Automation, University, China Direction Générale de l'Armement, France Shugen Ma Johann Dreo and Jean-Philippe Brunet Department of Robotics, Ritsumeikan University, Japan Thales, France

• A new framework for underwater object • Underwater constraints impose limited detection and localization is proposed. communications and detection. • An improved adaptive thresholding method to • UUVs maintain formation thanks to separate objects from background. neighborhood relative position. • Using a contour detection algorithm to find • UUVs don’t need to have unique identification objects’ contour for calculating the position of number. the objects. • Our formation algorithm is robust to UUV loss • The proposed framework can detect objects in the swarm, and to obstacles presence. UUVs swarm progression and calculate their position accurately. Underwater Object Detection and throught obstacles Localization

Tu3-1(3) 15:50-16:05 Tu3-1(4) 16:05-16:20

Kinematics Study on Pectoral Fins of Koi Carp Fluid Dynamics Experiments with a Passive by Digital Image Processing Robot in Regular Turbulence Lei Wang, Bo Liu, Min Xu, Jie Yang, Shiwu Zhang Department of Precision Machinery and Precision Instrumentation, University of Science and Gert Toming, Taavi Salumäe, Asko Ristolainen, Maarja Kruusmaa Technology of China, Hefei, Anhui, China Centre for Biorobotics, Tallinn University of Technology, Estonia • The experimental system for three Francesco Visentin, Otar Akanyeti Dept. of Computer Science, University of Verona, Italy dimensional kinematics is introduced to gain double perspective of pectoral fin of Koi Carp. • Passive fish robot in a Kármán vortex street. • Digital image processing is • Force and DPIV measurements. investigated to research the three- • Different positions from the cylinder both in dimensional kinematics of pectoral the downstream and lateral directions. fin of Koi Carp. • Acchieved the drag (force along the flow Three representative motions of • stream) reduction up to 42% with respect to pectoral fin: relaxation, expansion A motion cycle of the pectoral swimming in the uniform flow. and bending are extracted from the fin in retreating three dimensional grid graph.

Tu3-1(5) 16:20-16:35 Tu3-1(6) 16:35-16:50 Mapping and Visualizing Ancient Water Storage Hardware-in-the-Loop Verification for 3D Systems with an ROV – An Approach Based on Fusing Stationary Scans within a Particle Filter Obstacle Avoidance Algorithm of an William McVicker, Jeffrey Forrester, Timothy Gambin, Jane Lehr, Zoë J. Wood, and Underactuated Flat-Fish Type AUV Christopher M. Clark Saravanakumar Subramanian1 and Asokan Thondiyath3 • An offline stationary SLAM model Engineering Design, IIT Madras, India • Particle Filter used to determine ROV position for each Thomas George2 sonar scan with respect to all the others (no motion model) Mechanical Engineering, IIT Hyderabad, India • Data gathered from two expeditions to Malta and Sicily in 2011 and 2012 of underground cisterns and wells • HIL simulations play a major role for testing • Used VideoRay Pro 3 GTO underwater ROV and verifying the algorithms in real-time. equipped with a Tritech Micron Scanning Sonar and underwater sensors • 3D obstacle avoidance by multi-point potential • Autonomously generated 2D octree-based field method for an AUV is tested and verified. evidence grid maps of several cisterns. Raw Sonar Scan • SIL simulations are performed using MATLAB- Simulink and HIL test is executed using AUV on its development bench dSPACE. • Using HIL, the position and orientation errors are with in the acceptable limits. Fuse Individual Evidence Grid Maps using Particle Filter Final Map

41 IEEE ROBIO 2012 Conference Digest

Tu3-2: Biomimicking I Session Chairs: Liwei Shi and Alexey Gribovskiy

Room: Jade 2, 15:20-17:05, Tuesday, December 11, 2012

Tu3-2(1) 15:20-15:35 Tu3-2(2) 15:35-15:50

Development of a Venus Flytrap-inspired Robot Tadpole with a Novel Robotic Flytrap Biomimetic Wire-driven Propulsor Liwei Shi1, Shuxiang Guo1, 2, Hiroki Kudo1, and Kinji Asaka3 Baofeng Liao Center for Precisoin Engineering, Guangzhou Institute of Advanced Technology, 1Faculty of Engineering, Kagawa University, Japan Chinese Academy of Sicences, China 2College of Automation, Tianjin University of Technology, China 3Kansai Research Institute, AIST, Japan Zheng Li and Ruxu Du Dept. of Mechanical and Automation Engineering, the Chinese University of Hong Kong, China • To improve the detecting and grasping performances of Proximity sensor the previous developed robotic flytrap and evaluate its grasping ability, a new Venus flytrap-like robotic flytrap IPMC lobes • Introduce of wire-driven mechanism. was proposed. • Robot tadpole design. • We developed a prototype robotic flytrap and carried out trapping experiments by using the proximity sensor • Propulsion model and experiments. to imitate the trigger hairs of the actual Venus flytrap. • Discussions of experiment results and • The experimental results showed a good performance conclusion. of the robotic flytrap in mimicking the actual Venus Prototype robotic flytrap Motion modes of the flytrap. propulsor

Tu3-2(3) 15:50-16:05 Tu3-2(4) 16:05-16:20

CPG Parameter Search for a Biomimetic Robotic Numerical Analysis of Biomimetic Gannet Fish Based on Particle Swarm Optimization Impacting with Water during Plunge-diving

Zhengxing Wu, Junzhi Yu, and Min Tan Xingbang Yang, Tianmiao Wang, Jianhong Liang, State Key Laboratory of Management and Control for Complex Systems, Guocai Yao, Yang Chen, and Qi Shen Institute of Automation, Chinese Academy of Sciences, China Robotics Institute, Beihang University, China

• Reproduce the plunge-diving process of a • This paper addresses the parameter search issue of a Central Pattern biomimetic gannet by CFD method. Generator governed fishlike swimming. • Use the FLUENT 6.2 solver to solve the 3-D N- • A dynamic model of robotic fish swimming S equations. using Kane’s method is developed. • Track the air-water interface by the • A particle swarm optimization algorithm is VOF(Volume of Fluid) method. adopted to optimize the CPG parameters. • Present the impulse force and pressure distribution experienced by the plunge-diving The pressure distribution of a gannet under different initial water-entry plunge-diving gannet velocities.

Tu3-2(5) 16:20-16:35 Tu3-2(6) 16:35-16:50 A Bio-mimetic Fingertip That Detects Force and Building a Safe Robot for Behavioral Vibration Modalities and its Application to Experiments Surface Identification Alexey Gribovskiy and Francesco Mondada K. V. D. S. Chathuranga, Van Anh Ho and Robotic Systems Laboratory, EPFL, Switzerland Shinichi Hirai José Halloy Department of Robotics, Ritsumeikan University, Japan LIED, Université Diderot Paris VII, France Jean-Louis Deneubourg • Design, fabrication and data processing Unit of Social Ecology, Université Libre de Bruxelles, Belgium method of a biomimetic soft tactile fingertip for • Guidelines to build a safe robot for behavioral humanoid robots is presented. biology studies • This soft fingertip has accelerometers and • Safety elements of the PoulBot robot force sensors in two layers of polyurethane • How on-board and off-board sensors can be rubber and able to detect micro-vibration and combined to successfully detect and avoid force modalities. abnormal situations • The fingertip is evaluated for the ability to differentiating surface profiles Overview of the PoulBot robot

Surface identification

42 IEEE ROBIO 2012 Conference Digest

Tu3-3: Medical & Surgery Robots I Session Chairs: Med Amine Laribi and Daniel Bernd Roppenecker

Room: Jade 3, 15:20-17:05, Tuesday, December 11, 2012

Tu3-3(1) 15:20-15:35 Tu3-3(2) 15:35-15:50

Modeling of Flexible Arm with Triangular Notches Development of a Maxillofacial Multi-arm for Applications in Single Port Abdominal Surgery Surgery Robot Wei Dong, Wenlong Yang, Dawei Han, and Zhijiang Du Xiangzhan Kong, Xingguang Duan and Yonggui Wang State Key Laboratory of Robotics and System, Intelligent Robotics Institute, Beijing Institute of Technology, China Harbin Institute of Technology, Harbin, Heilongjiang, China • The Maxillofacial Multi-arm Surgery Robot A new flexible arm with triangular notches • System (MMSRS) was introduced . is proposed for the application in single port abdominal surgery. • The mechanical structure and control system of the Maxillofacial Multi-arm Surgery Robot The mechanism model of the flexible arm is • (MMSR) was designed. presented based on the Timoshenko’s beam theory. • Forward kinematics and inverse kinematics of the robot were analyzed. •Experimental results showed a good The solid model and comparison results agreement with the proposed theoretical • simulation and experiments were done to test mechanics model. the performance of robotic system. The Maxillofacial Multi-arm Surgery Robot

Tu3-3(3) 15:50-16:05 Tu3-3(4) 16:05-16:20

A design of slave surgical robot based on Statistical Atlas Based Registration and Planning for motion capture Ablating Bone Tumors in Minimally Invasive Interventions Xin Kang1, PhD; Hongliang Ren2*, PhD; Jing Li3, MD, PhD; Wai-Pan Yau4, MD 1Children’s National Medical Center, Washington D.C, USA M. A. LARIBI, T. RIVIERE, M. ARSICAULT and S. ZEGHLOUL 2* Dept of Bioengineering, National University of Singapore, Singapore PPRIME Institute, DGMSC CNRS – Poitiers University – ENSMA-UPR 3346, 3Dept of Orthopaedics, Xijing Hospital and The Fourth Military Medical Univ.,China. FRANCE 4Dept of Orthopaedics and Traumatology, The University of Hong Kong, Hong Kong

• A slave robot architecture for minimally • Minimally invasiveness using sparse invasive surgery (MIS) compact and close to radiograph & statistical atlas surgical gesture was proposed • X-ray to Altas registration to get • A new design methodology using motion patient-specific model capture and biomechanics expertise is applied • Ablation optimization for treating large • A mathematical formulation of the spherical tumors mechanism workspace is used in the synthesis • Benefits for patients & clinicians. Atlas-based registration and planning system for tumor ablation problem • Exceptional for high-risk patients such as • An extensive database surgical measurements young females and children is established

Tu3-3(5) 16:20-16:35 Tu3-3(6) 16:35-16:50

Interdisciplinary development of a Single-Port Towards the development of a robust model for Robot estimating wrist torque at different wrist angles Daniel B. Roppenecker and Tim C. Lueth Micro technology and medical device technology, TUM Munich, Germany Alexander Meining Mojgan Tavakolan and Carlo Menon MENRVA Research Group, SFU, Burnaby, Canada II. Medical Department, Klinikum rechts der Isar, TUM Munich, Germany Gerald Horst and Heinz Ulbrich Institute of applied mechanics, TUM Munich, Germany • A model is sought that is unaffected by varying • Single-Port Robot for minimal invasive arm postures and joint angles. Endoscopic Submucosa Dissection. • Obtained results for overcoming this problem • Using of selective laser sintering and the are presented. biokompatible material PA2200. • Training a linear regression model with data • First realized Prototype based on clinical from varied joint angles increased the requirements. accuracy. Trained With Different Joint • Guidance of flexible instruments up to Angles Ø 2.8 mm. • Manually controlled by cables. Realized Prototype

43 IEEE ROBIO 2012 Conference Digest

Tu3-4: Mobile Robots I Session Chairs: Rene Matthias and Jin-hui Zhu

Room: Jade 8, 15:20-17:05, Tuesday, December 11, 2012

Tu3-4(1) 15:20-15:35 Tu3-4(2) 15:35-15:50 Bond Graph Modeling of an Over-actuated Intelligent Autonomous Vehicle with Decoupled Introducing a Standardized Electronics Architecture for Multi-Modular Self-Reconfigurable Mobile Robots Steering Wheel Tarun Kumar Bera Mechanical Engg. Deptt, Thapar University, Patiala, India Arun Kumar Samantaray Rene Matthias and Heinz Wörn IPR, Karlsruhe Institute of Technology (KIT), Germany Mechanical Engg. Deptt, IIT, Kharagpur, India Rochdi Merzouki and Belkacem Ould-Bouamama LAGIS, USTL, Lille, France • New architecture for multi-modular • This article deals with modeling and control of A figure here self-reconfigurable mobile robots an intelligent autonomous vehicle. • High-speed Ethernet bus for inter- • The bond graph model of IAV is developed in a organism communication modular and hierarchical modeling • Accumulating bandwidth, arbitrary environment. topologies, standard components • The Ackerman steering concept was used in • Decomposition of modules into the four wheel steering motor controller. RobuTAINeR’s prototype for various submodules • The controller based on pure pursuit algorithm 40 feet transporter container and a PI controller are used as hybrid control.

Tu3-4(3) 15:50-16:05 Tu3-4(4) 16:05-16:20

Design of a wheel-type mobile robot for rough Passive scattering transform bilateral teleoperation for terrain an Internet-based mobile robot Jin-hui Zhu, Bin-bin Li, Yi-zhen Li, Hua-qing Min School of Software Engineering, South China University of Technology, China • The random time delay brings great challenges to the controller design for the Internet-based teleoperation system, the worst case is that it can destabilize the Internet-based teleoperation system. This paper first gives Russell Thomas James a brief historical preview of bilateral passive , and then a direct passive scattering transform method, extending the passive bilateral control method to the Internet-based random time delay DAGU Hi-Tech Electronic Co., Ltd,China teleoperation system, is advanced to guarantee the teleoperation system’s stability with any non-symmetric random network time delay. Finally, a virtual master-slave manipulator bilateral control scheme is designed for • iPR has two different float suspension mobile robot teleoperation based on the passive scattering transformation method, and simulations are carried out to verify the results at the end of this paper. mechanisms with six wheels to handle rougher A figure here terrain. • Even if a tire is damaged iRP can continue on the remaining 5 wheels. • iRP’s six wheels remain on the ground even if any single wheel is raised. • iRP can be applied to security, factory inspection, family inspection and product shopping guiding. iRP robot

Tu3-4(5) 16:20-16:35

Design of Vertical Climbing Robot with Compliant Foot Yanwei Liu, Chongyang HU, Xuan Wu and Yongjie Zhang University of Science and Technology of China, China Tao Mei and Shaoming Sun Hefei Institute of Physical Science, Chinese Academy of Sciences,China

• This paper presents a vertical climbing robot which uses inchworm-like motion and bidirectional claws. • A single motor drives the crank-slider mechanism to imitate the inchworm motion. • A bidirectional claw is designed and analyzed The robot is 12cm long, 30g to make the attachment reliable. and capable of climbing vertical • The experiment shows that this robot is more pearl brocade plate, softwood energy efficient than other vertical climbing and other soft or rough surface robots. with 0.5 body weight as payload.

44 IEEE ROBIO 2012 Conference Digest

Tu3-5: Grasping and Manipulation III Session Chairs: Heping Chen and De Xu

Room: Jade 5, 15:20-17:05, Tuesday, December 11, 2012

Tu3-5(1) 15:20-15:35 Tu3-5(2) 15:35-15:50

Integrative Joint for Space Manipulator Dynamic Simulation and Analysis for Bolt and Nut Mating of Dual Arm Robot

Yu He, Hui Li, Zhihong Jiang, Que Dong, Fuhai Zhang, Lei Hua, Yili Fu, and Bin Guo Bo Wei, and Qiang Huang Robotics Institute, Harbin Institute of Technology, China School of Mechatronical Engineering, Beijing Institute of Technology, China

• We have finished the kinematic analysis of • High-cohesion joint is designed using closed chain system and the arms. Integrative method • We analyze the kinematic constraint relations • Accurate dual-redundancy detection for for bolt and nut mating of dual arm robot. motor and joint positions is realized by • By using simulation tool (ADAMS software), 3D two same resolvers dynamic simulation is performed.

• Problem of joint heat dissipation in Mechanical structure of • Simulation results show that this method can vacuum is solved by thermal control and integrative joint complete bolt and nut mating smoothly. isothermal design 3-D model of the dual arm robot

Tu3-5(3) 15:50-16:05 Tu3-5(4) 16:05-16:20 An Algorithm for Object Manipulation with Obstacle Modeling for Manipulator Using Three-fingered Robotic Hands Qin Zhang, Zhi Chen Iterative Least Square (ILS) and Iterative Closest School of Mechanical & Automotive Engineering,South China University of Point (ICP) Base on Kinect Technology,Guangzhou China Kamiya Yoshitsugu Wantana Sukmanee, Miti Ruchanurucks, and Panjawee Rakprayoon Department of Electrical Engineering, Kasetsart University, Bangkok, Thailand Department of Mechanical Systems Engineering, Kanazawa University, Kanazawa, Japan • This paper present a method to distinguish • The mechanical model of the multi- between a manipulator and its surroundings fingered hand using a depth sensor. • Manipulation algorithm used for the • The depth sensor used is Kinect. multi-fingered hand • Error in the latter calibration is minimized by • Simulation and verify the feasibility using Iterative least square (ILS) method and and performance of the proposed Iterative closest point (ICP) method. algorithm • Experiment shows that an overlapping Point cloud and 3D model between point cloud and 3D model of manipulator.

45 IEEE ROBIO 2012 Conference Digest

Tu3-6: Machine Learning II Session Chairs: Manabu Gouko and Dan Xu

Room: Jade 9, 15:20-17:05, Tuesday, December 11, 2012

Tu3-6(1) 15:20-15:35 Tu3-6(2) 15:35-15:50

A Novel Formalization of Affordance Model for An Improved Algorithm for Reducing the Influence of Muscle Fatigue in sEMG based HMI Yi Zhang,Xinli Xu and Yuan Luo Chang’an Yi, Huaqing Min, Ronghua Luo, Zhipeng Zhong and Research Center of Intelligent System and Robot, Xiaowen Shen Chongqing University of Posts and Telecommunications, China Computer Science and Engineering, South China University of Technology, Huosheng Hu China Computer Science & Electronic Engineering, University of Essex, UK Huiyu Zhou • A new 4-tuple formalization for ECIT,Queen’s University Belfast, UK Affordance learning Precondition • HMI based on sEMG signals suffers • A 3-level Affordance model based on Action from poor reliability and robustness in analysis function long-term operations. Postcondition • An Affordance based robot control • This problem is mainly caused by architecture Affordance Model muscle fatigue. • A novel method is proposed and it is A subject is driving a Four units in our formalization effectively diminish the influence of wheelchair using designed muscle fatigue in long-term operations. sEMG based HMI

Tu3-6(3) 15:50-16:05 Tu3-6(4) 16:05-16:20

A Spectral Clustering Method Combining Path Dynamic Real-time Dynamic Gesture Recognition System with Density Based on Depth Perception for a,b b a Dan Xu1, Yen-Lun Chen1, Chuan Lin1, Xin Kong1 and Xinyu Wu1,2 Hongwei Xu , Jiafeng He , Qing He Dewen Zenga, Guan Guana, Bin Lenga, and Weimin Zhenga 1 Shenzhen Institutes of Advanced Technology, Chinese Academy of Sciences a 2 Department of Mechanical and Automation Engineering, CUHK, HongKong Guangzhou Institutes of Advanced Technology, Chinese Academy of Sciences bSchool of Information Engineering, Guangdong University of Technology, China

• This paper proposes a fast spectral clustering method that utilizes the density of dataset to limit the scope of paths instead of finding all the paths. • Some experiments are conducted to compare with the path-based algorithm. Results on USPS by Rand Index

Tu3-6(5) 16:20-16:35 Tu3-6(6) 16:35-16:50

On Learning Coordination Among Soccer Reinforcement learning for discernment Agents behavior acquisition Ali Raza, Usman Sharif and Sajjad Haider Manabu Gouko Dept. of Mechanical Engineering and Intelligent Systems, Faculty of Lab, Faculty of Computer Science Engineering, Tohoku Gakuin University, Japan Institute of Business Administration Karachi, Pakistan Yuichi Kobayashi Dept. of Mechanical Engineering, Shizuoka University, Japan Chyon Hae Kim • The paper applies Neural networks to learn Honda Research Institute Japan, Japan coordination between two soccer agents within RoboCup Soccer 3D Simulation domain. • We propose an active perception model that • Neural networks learn by observing log files of autonomously learns discernment behaviors the matches played by one of the better teams. for a robot. • Knowledge is extracted from log files by • After the learning, the robot acquired defining events. different behaviors according to each object. • The goal is to predict the next position of the • The model effectively established intelligent robot based on relevant variables. strategies according to the relation between Proposed model object features and robot's configuration.

46 IEEE ROBIO 2012 Conference Digest

Tu3-7: SLAM II Session Chairs: Han Wang and Xinzheng Zhang

Room: Jade 7, 15:20-17:05, Tuesday, December 11, 2012

Tu3-7(1) 15:20-15:35 Tu3-7(2) 15:35-15:50

Visual Odometry using RGB-D Camera on Estimating Terrain Elevation Maps from Ceiling Vision Sparse and Uncertain Multi-Sensor Data Han Wang, Wei Mou, Hendra Suratno, Gerald Seet, Maohai Li, Dominik Belter, Przemyslaw Labecki, Piotr Skrzypczynski M.W.S. Lau and Danwei Wang Institute of Control and Information Engineering School of Electrical & Electronic Engineering, Poznan University of Technology, Poland Nanyang Techonological University, Singapore

• A Principal direction detection approach using • This paper addresses the issues of terrain RGB-D camera for ceiling vision is introduced. modelling for the purpose of motion planning in • Lines from ceiling are used as reference for an insect-like walking robot.. current robot’s heading direction. • We extend the elevation grid update rules by • The principal direction is used to achieve incorporating formal treatment of the spatial higher accuracy for visual odometry estimation uncertainty. based on ICP. • Moreover, we describe a novel method for filling- • The proposed approach can be operated in System flow chart in missing areas of the elevation grid using real-time and perform well even with camera Gaussian-based approximation. disturbance.

Tu3-7(3) 15:50-16:05 Tu3-7(4) 16:05-16:20

A New Localization Method for Mobile Robots by Data Localization of a simple robot with low computa- Fusion of Vision Sensor Data and Motion Sensor Data tional power using a single short range sensor Tae-jae Lee, Wook Bahn, Byung-moon Jang Seoul National University, Korea Ho-Jeong Song Demetris Stavrou and Christos Panayiotou Hagisonic Co., Korea ECE Department, University of Cyprus, Cyprus Dong-il “Dan” Cho Seoul National University, Korea • A new localization method by the data fusion of vision sensor and motion • Monte Carlo Localization is used to track sensor measurements is proposed. multiple potential poses of the robot A figure here • Developed system estimates the orientation, position and velocity of a • The robot is equipped with a single short range mobile robot. infrared sensor • The developed system uses data from a camera, robot wheel encoders, • Minimal number of particles is used due to the and accelerometer, and a gyroscope. low computational capabilities of the robot • The experiments are performed to verify the localization performance in • Localization achieved in a static environment with random dynamic events Localization Performance several motion scenarios, including a normal operation mode, a wheel slip Blue – Ground Truth mode, and a kidnap mode. Red – Localization Algorithm

Tu3-7(5) 16:20-16:35 Tu3-7(6) 16:35-16:50

A New Algorithm for Robot Localization Using Keypoints Based Laser Scan Matching - A Monocular Vision and Inertia/Odometry Sensors Robust Approach R. Ray, D. Banerji, S. Nandy and S. N. Shome Kai Wang, Yunhui Liu, and Luyang Li CSIR-Central Mechanical Engineering Research Institute, Durgapur, India The Chinese University of Hong Kong, Hong Kong • Keypoints are extracted from the Image obtained from LRF data using Harris • Fusion of the monocular vision and corner detection method. inertia/odometry sensors for estimating • Robot pose is estimated by matching the key points of two successive Image the robot position. frames using RANSAC method. • Fusion algorithm based on adaptive • This method can be implemented for high rotation and translation effectively estimation. compared to the golden method like ICP. • Proved convergence and robustness. • The proposed technique takes higher computational time, but more Robust to noisy data. • GUP implement for the real-time performance. • The increased computational time can be justified if the environment is cluttered and/or the sensor data is noisy.

47 IEEE ROBIO 2012 Conference Digest

Tu3-7: SLAM II (cont.) Session Chairs: Han Wang and Xinzheng Zhang

Room: Jade 7, 15:20-17:05, Tuesday, December 11, 2012

Tu3-7(7) 16:50-17:05 Incremental Human Motion Map System and Human Walking Behavior Representation in Indoor Environment Tetsuya Wada, Zhidong Wang, Yuji Ogawa, Dept. of Advanced Robotics, Chiba Institute of Technology, Japan Yasuhisa Hirata and Kazuhiro Kosuge System Robotics Lab., Tohoku University, Japan • A human motion map building method incorporating SLAM and human estimation algorithm in dynamic environment is proposed. • Dynamic and static information are extracted from maps generated by a SLAM procedure implemented. • Interpolation algorithms are designed and applied to build human motion maps. Human Motion Map

48 Wednesday, December 12, 2012

We1-1 Underwater Robots II We1-2 Biomimicking II We1-3 Medical & Surgery Robots II We1-4 Mobile Robots II We1-5 Aerial Robots I We1-6 Smart Structures, Materials, Actuators II We1-7 Service Robots II We2-1 Micro/nano Devices & Robots We2-2 Biomimicking III We2-3 Medical & Surgery Robots III We2-4 Human Robot Interaction I We2-5 Aerial Robots II We2-6 Smart Structures, Materials, Actuators III We2-7 Service Robots III We3-1 Intelligent Systems I We3-2 Biologically Inspired Robotics III We3-3 Sensor Networks We3-4 Human Robot Interaction II We3-5 Rehabilitation Robots I We3-6 Machine Learning III We3-7 Man Machine Interface I IEEE ROBIO 2012 Conference Digest

We1-1: Underwater Robots II Session Chairs: Young-sun Ryuh and Shugen Ma

Room: Jade 1, 10:20-12:05, Wednesday, December 12, 2012

We1-1(1) 10:20-10:35 We1-1(2) 10:35-10:50

Analysis of Swimming Patterns of Diving Characteristics Evaluation of the Vertical Motion Beetles incorporation Robotics of a Spherical Underwater Robot Chunfeng Yue1, Shuxiang Guo2,3, Liwei Shi2 and Juan Du3 Hee Joong KIM, Jihong LEE 1. Graduate School of Engineering, Kagawa University, Japan 2. Faculty of Engineering, Kagawa University, Japan Department of Mechatronics, Chungnam National University, KOREA 3.College of Automation, Harbin Engineering University, China

• Biomimetic Technology( Diving beetle and • Evaluated the vertical movement Articulated Underwater Robot) characteristics.

• Applying Swimming patterns of a diving θ1 θ2 • Estimated the relative parameters of vertical beetle to a robot movement and carried out some simplifications • Comparison of Trajectory and Joint Angle to calculate the moving situation. changes Between a Diving Beetle and a • Pressure sensor is used into the closed-loop Robot control system. • Confirmation of high similarities in Trajectory • Carried out some experiments to verify the The explosive view of the and Joint angle changes feasibility and performance of the proposed spherical underwater robot vertical control system.

We1-1(3) 10:50-11:05 We1-1(4) 11:05-11:20

Effect Analysis of Chordwise Flexibility on Propulsion Bio-Inspired Multi-Robot Communication Performance of Oscillating Pectoral Foils through Behavior Recognition

Lige Zhang, Shusheng Bi, Yueri Cai, Chuanmeng Niu, and Michael Novitzky, Charles Pippin, Thomas R. Collins, Tucker R. Hongwei Ma Balch, and Michael E. West Robotics Institute, Beihang University, China College of Computing, Georgia Institute of Technology, USA

• The study on the flexibility to promote the • AUV/ASV team in a communication denied performance impact is introduced. environment A figure here • The chordwise flexible defromation model is • AUV uses behavior similar to honeybee built. “waggle dance” to show location of MLO • Impact of the chordwise flexibility on the • Several trials demonstrate feasibility of oscillating propulsion is carried out by behaviors for communication hydrodynamic experiments. Experimental platform For figure caption (optional), use arial narrow 20pt font

We1-1(5) 11:20-11:35

Multi-agent Control and Implementation of Bio-inspired Underwater Robots for Mariculture Monitoring and Control Young-sun Ryuh Biomimetic robots research group, Korea Institute of Industrial Technology, Korea Sth. Jeon-il Moon Daegu Gyeongbuk Institute of Science & Technology

• Marine environment like as mariculture is not feasible for human beings to run it by themselves and a single robot can hardly collect all information required, and all of monitoring work will be stopped in case of the robot's breakdown. • This paper proposes an underwater swarm robot system with Petri Net in which each of multiple agent robots operates on its own according to its own private environment being able to locate itself based on robot buoys

49 IEEE ROBIO 2012 Conference Digest

We1-2: Biomimicking II Session Chairs: Shuxiang Guo and Yili Fu

Room: Jade 2, 10:20-12:05, Wednesday, December 12, 2012

We1-2(1) 10:20-10:35 We1-2(2) 10:35-10:50

A Rat-like Robot WR-5 A New Design and Analysis of Compliant Spine for Animal Behavior Research Mechanism for Caterpillar Climbing Robot Qing Shi, Hiroyuki Ishii, Hikaru Sugita, WANG Wei and XIE Dawei Shinichi Kinoshita, Zhuohua Lin, Atsuo Takanishi, Robotics Institute, School of Mechanical Engineering and Satoshi Okabayashi, Naritoshi Iida, Hiroshi Kimura Automation, Beihang University, China Waseda University, Japan

• Capable of mimicking rat behavior • Spine attachment lowers energy consuming and is suitable for various surfaces. • Endowed with almost the same motion A figure here performance as living mature rats • The trajectory of spines is discussed. • Achieve to inspire behaviors of rats • The model of the compliant driving mechanism robot is built for specific design. • Better controllability and higher reproducibility than conventional animal experiments • The movement of a spine is simulated ,whose results verify the feasibility of our design . • Able to make rats feel fear by means of Rat-like robot WR-5 and a stuffed performing mounting action rat made from a mature rat

We1-2(3) 10:50-11:05 We1-2(4) 11:05-11:20

Design and Kinematics Simulation for Bionic 4D Simulation and Control System for Life Crank-slider Mechanism of Jumping Robot Science Automation

Yanfei Lia), Steffen Jungingerb), Norbert Stollb) and Kerstin Thurowa) Zhao Kai, Yin Junmao, Chen Diansheng and Wang Tianmiao a. Center for Life Science Automation (celisca), Germany Robotics Institute, Beihang University, China b. Institute of Automation, University of Rostock , Germany • Observe the locust morphology and analyze the posture of hindlimbs at take-off phase. • The paper presents a 4D Simulation and Control System to realize an advanced virtual • Propose a crank-slider mechanism to mimic the dynamic simulation. locust hindlimbs. • The system blends functions of PMS and 3DCreate, realizes a real-time and visual • Design a robot which employs the crank-slider mechanism and segment-gear system. dynamic simulation. Working process of the 4D • It makes PMS in Life Science Automation Simulation and Control System. The 3D model of the • Verify its jumping performance by kinematic flexible, releases designers from heavy and jumping robot modeling. boring assays.

We1-2(5) 11:20-11:35 We1-2(6) 11:35-11:50

Studies of motor synergies in generating optimal Development and Depth Control of a Robotic goal-directed movements in human-like Fish Mimicking Cownose Ray Kin Chung Denny Fu, Yutaka Nakamura, Tomoyuki Yamamoto, Chuanmeng Niu, Lige Zhang, Shusheng Bi and Yueri Cai and Hiroshi Ishiguro Robotics Institute, Beihang University, China Department of Systems Innovation, Osaka University, Japan • Inspiration from human motor control studies, control is simplified by utilizing motor • A robotic fish mimicking cownose ray is synergies. A figure here developed. • Motor synergies possess “goal-related” and • A fuzzy depth control method is presented. “goal-non-related” properties were studied. • Swimming experiments are carried out to verify • Reaching tasks can be achieved by utilizing the effectiveness of the depth control method. motor synergies possess property of “achieving goal optimally”.

• Reaching tasks can also be achieved by Easier to control by fewer number Prototype of Robo-ray III utilizing motor synergies possess property of of motor synergies “energy efficiency”.

50 IEEE ROBIO 2012 Conference Digest

We1-3: Medical & Surgery Robots II Session Chairs: Ran Zhou and Jason Gu

Room: Jade 3, 10:20-12:05, Wednesday, December 12, 2012

We1-3(1) 10:20-10:35 We1-3(2) 10:35-10:50

Endoscopes Shape Reconstruction Based on Wireless Capsule Endoscopy Video Electromagnetic Localization and Curve Fitting Automatic Segmentation Ran Zhou, Baopu Li, Zhe Sun, Chao Hu and Max Q.-H Meng Xiaoxiao Li, Baopu Li, Shuang Song, Chao Hu Shenzhen Institutes of Advanced Technology, Chinese Academy of Sciences Shenzhen Institutes of Advanced Technology, Chinese Academy of Sciences, Shenzhen, China

• Proposed to reduce the number of • A novel and efficient algorithm is proposed sensors to improve the design of in this paper to help clinicians segment the colonoscope. WCE video automatically • Localize the sensors based on • Denote valid regions in WCE frames based electromagnetic method. on color and texture features • Reconstruct the endoscopes shape • The proposed two level WCE video using a curve fitting method. segmentation method • Experiments and conclusions. Example of classification and segmentation results

We1-3(3) 10:50-11:05 We1-3(4) 11:05-11:20

Design and Optimization of a Parallel Guidance Ankle-Angle Analysis of Drop-Foot Patient in Device for Minimal Invasive Spinal Surgery Walking with FES Peng Gao, Ying Hu, Haiyang Jin, Peng Zhang, Jianwei Zhang

Ningjia Yang, Feng Duan, Sicen Liu and Lili Dai Institutes of Advanced Technology, Chinese Academy of Sciences, Shenzhen, China Department of Automation, Nankai University, P. R. China The Chinese University of Hong Kong, Hong Kong, China

• This study aims to quantify the effect • Developed a guidance device, which is based of FES on hemiplegic drop-foot on 3-RPS parallel platform. patients’ movements. • Proposed dynamics indicators Based on the • Using motion capture system to frequency response of the quadratic response measure subjects’ walking surface . performance. • A integrated optimization method based on the • Comparing the measurements with kinematics and dynamics of multi-objective and without FES. optimization is proposed. Subject walking with FES The figure of minimally invasive spinal surgery guide device

We1-3(5) 11:20-11:35 We1-3(6) 11:35-11:50

Posture Study for Self-training System of Kinematics, control and workspace analysis of a bowden wire Patient Transfer actuated manipulator for minimally invasive single-port surgery Zhifeng Huang, Ayanori Nagata, Taiki Ogata, and Jun Ota Research into Artifacts, Center for Engineering(RACE), The University S. Can, B. Jensen, E. Dean-Leon C. Staub, A. Knoll of Tokyo, Japan Robotics and Embedded Systems, Technical University Munich A. Fiolka, A. Schneider, A. Meining H. Feussner Masako Kanai-Pak, Jukai Maeda, Yasuko Kitajima, Klinikum rechts der Isar, Technical University Munich Mitsuhiro Nakamura, and Kyoko Aida Faculty of Nursing, Tokyo Ariake University of Medical and Health • Robot platform for laparoscopic single-port surgery Sciences, Japan comprising two highly versatile manipulators and a Noriaki Kuwahara double-bending telescope Department of Advanced Fibro-Science, Kyoto Institute of Technology • The platform enables the introduction of flexible endoscopic instruments through hollow manipulators • The purpose of this research is to construct that are actuated by bowden wires in 2m distance a self training system of patient transfer. • Kinematics of the manipulator and the implemented velocity-based task-space control in a simulation are • Measurement method of posture was based presented in this work on range camera Kinect and color marks. • Results of a pick-and-place scenario for evaluation of the kinematics as well as the workspace analysis and Highly Versatile Single-Port • Bayes minimum error decision was perform manipulability of the manipulator are presented System to classify the posture as right or wrong. Sample image of patient transfer

51 IEEE ROBIO 2012 Conference Digest

We1-4: Mobile Robots II Session Chairs: Yunhui Liu and Kazuhito Yokoi

Room: Jade 8, 10:20-12:05, Wednesday, December 12, 2012

We1-4(1) 10:20-10:35 We1-4(2) 10:35-10:50

Variable-resolution Velocity Roadmap Generation Mobile Robot Action Based on QR code Considering Safety Constraints for Mobile Robots Identification Jingyu Xiang, Yuichi Tazaki, Tatsuyta Suzuki and B. Levedahl Wenyu Li Dept. of Mech. Sci. and Eng. Department of Intelligence Science, Nankai University, P. R. China Graduate School of Engineering, Nagoya University, Japan Feng Duan Department of Automation, Nankai University, P. R. China • Reuse a conventional roadmap and add a • This study aims to develop the velocity space to the roadmap. application of QR code identification in • Each node in the proposed roadmap consists robot action. a fixed position and a velocity regions. • A fast identification approach for QR • Use variable-resolution partition for dividing code in the indoor environment was velocity space. proposed. • Connectivity of nodes is based on safety and • A feedback control planar mobile robot specification constraints. was used as a platform to verify the Robot moved and grasped the object based on the QR code. • Inputs can be found in planning phase. An environment and a proposed method. roadmap

We1-4(3) 10:50-11:05 We1-4(4) 11:05-11:20

Event-based Particle Filtering Ball Dynamics Simulation on OpenHRP3 for Robot Self-Localization David Weikersdorfer Intelligent Autonomous Systems, TU Munich, Germany Rafael Cisneros*^, Eiichi Yoshida^, Kazuhito Yokoi^* * Systems and Information Engineering, University of Tsukuba, Japan Jörg Conradt ^ CNRS-AIST JRL (Joint Robotics Laboratory), UMI 3218/CRT, AIST, Japan Neuroscientific System Theory, TU Munich, Germany

• We present a particle filter tracking algorithm • An integrated software platform for robot for event-based image sensors. simulations and software developments, A figure here • Our algorithm works with sub-millisecond OpenHRP3, is improved to achieve realistic accuracy directly on singular highly ambiguous ball dynamics simulations. events. • It treats the ball as a sphere when detecting its • Evaluation in a robot self-localization task with collision against objects modeled as meshes. a simulation environment and experimental • This algorithm represents an efficient method recorded data. to calculate the collision data: contact points Collision between the ball and just one triangle of the mesh and force reaction vectors.

We1-4(5) 11:20-11:35 We1-4(6) 11:35-11:50

A Globally Asymptotic Attitude Estimation Using A Real-time Method for Detecting Sharp Images Complementary Filtering in Visual Navigation Junjun Wu1, Yisheng Guan1, Manjia Su1 and Hong Zhang1,2 BENZIANE Lotfi and Benallegue Abdelaziz 1 School of Mechanical and Automotive Engineering, El Hadri Abdelhafid South China University of Technology, China Versailles Engineering Systems Laboratory, UVSQ, FRANCE 2 Department of Computing Science, University of Alberta, Canada

• We propose a simple and efficient globally  A real-time method to filter the blurred images asymptotic attitude estimation solution. for improving the robustness of visual navigation to motion blur is presented. • Complementary linear filters are combined with  Based on the visual blur feature and the robotic TRIAD algorithm. motion feature, the method works on-line in a • Lyapunov analysis results for the proposed no-reference detection framework. direct and passive second order filters are  The two-step detection procedure is addressed. Proposed Attitude and highly reliable. • The proposed solution includes the estimation Heading Reference System  The method is more efficient than existing of gyro-rate bias . structure methods in terms of real-time performance.

52 IEEE ROBIO 2012 Conference Digest

We1-5: Aerial Robots I Session Chairs: Gerasimos Rigatos and Michael Wang

Room: Jade 5, 10:20-12:05, Wednesday, December 12, 2012

We1-5(1) 10:20-10:35 We1-5(2) 10:35-10:50

Sensor fusion for UAV navigation based on Monocular Visual SLAM for Small UAVs in Derivative-free nonlinear Kalman Filtering GPS-denied Environments Gerasimos G. Rigatos Unit of Industrial Automation, Industrial Systems Institute, Greece Chaolei Wang, Tianmiao Wang, Jianhong Liang, Yang Chen, Yicheng Zhang and Cong Wang • The Derivative-free nonlinear Robotics Institute, Beihang University, China Kalman Filter (DKF) is compared to other nonlinear estimators (EKF, • EKF based monocular visual SLAM UKF, PF), in the problem of sensor in GPS-denied environments for fusion-based nonlinear control of small UAVs unmanned aerial vehicles. • SIFT based landmark detection and match • It is shown that the Derivative-free nonlinear Kalman Filter is faster • Homography based visual motion than the other nonlinear estimation estimation • Inverse depth landmark initialization algorithms while its accuracy of SLAM simulation environment estimation is also satisfactory. and reparameterization

We1-5(3) 10:50-11:05 We1-5(4) 11:05-11:20

Self-tuning of 2-axis autopilot Structural Target Recognition Algorithm for for small aeroplanes and aerial Visual Guidance of Small Unmanned Helicopters robots Yi Zhou, Tianmiao Wang, Jianhong Liang, Drago Matko, Tine Tomažič and Rok Markovič Chaolei Wang, Yicheng Zhang Robotic Institute, Beihang University, China

• A vision based guidance approach for small unmanned helicopters. A figure here • A Hu invariant moments based Structural target recognition method for visual acquisition of the target. • A HSL model based filter is designed to eliminate the noises from the image. • A visual guidance model is designed and Flight experiment implemented to locate the target.

We1-5(5) 11:20-11:35

Cooperative vSLAM Based on UAV Application

Xiaodong Li Informatics and Systems Engineering, Cranfield University, UK Nabil Aouf Informatics and Systems Engineering, Cranfield University, UK • A strategy of cooperative visual SLAM was proposed to enhance estimation accuracy of UAV trajectory. • System dynamic processing and stereo camera observation models for UAVs were well studied and presented. • Extended Information Filter was utilized based on its advantage for multiple platforms data fusion. • Covariance Intersection algorithm was implemented to achieve distributed and decentralized data fusion. • Performance was depicted intuitively with encouragement to the further research.

53 IEEE ROBIO 2012 Conference Digest

We1-6: Smart Structures, Materials, Actuators II Session Chairs: Masamitsu Kurisu and Xinkai Chen Room: Jade 9, 10:20-12:05, Wednesday, December 12, 2012

We1-6(1) 10:20-10:35 We1-6(2) 10:35-10:50

An Omnidirectional Mobile Robot: Concept and Fault Detection and Diagnosis based on Sparse Analysis Representation Classification (SRC) Shugen Ma and Chao Ren Department of Robotics, Ritsumeikan University, Japan Lijun Wu, Xiaogang Chen, Yi Peng, Qixiang Ye and Jianbin Jiao Changlong Ye School of Electronics and Communication Engineering, Department of Mechatronics Engineering, Shenyang University of Chinese Academy of Sciences, China Aerospace University, China • A novel omnidirectional wheel mechanism • Conducting the dictionary with all fault patterns based on sliced ball structure, referred to as A figure here training samples for SRC. A figure here MY wheels-II, has been proposed. • Computing the sparse reconstruction • The problem of wheel angular velocity coefficients and residuals when a test sample fluctuations caused by the specific mechanical comes. structure is studied. • Selecting the fault pattern which has the • The methods used for determining the OSF are smallest reconstruction residual as the FDD Framework of FDD based discussed briefly. Prototype Platform Based on result. on SRC MY Wheels-II

We1-6(3) 10:50-11:05 We1-6(4) 11:05-11:20

Development of Hopping Mechanism using Nonlinear Contact and Impact of Revolute Joints Permanent Magnets for A Tiny Asteroid Rover - Estimation of Hopping Velocity - Rongqiang Liu, Jing Zhang, Hongwei Guo, and Zongquan Deng Masamitsu Kurisu School of Mechatronics Engineering, Harbin Institute of Technology, Department of Mechanical Engineering, Tokyo Denki University, Japan Harbin, China

• Correction contact model is accurate to • Author has proposed the hopping mechanism calculate the contact deformation of revolute using permanent magnets for a tiny asteroid joint. exploration rover. • The global displacement model of revolute joint • The mechanism utilizes for hopping the impact is built based on the simplified curved beam. force generated by adsorption of the magnets. • The motion simulation of revolute joint is • This paper presents an estimation method of basically agree with the experimental results. the hopping velocity just after the impact. Revolute joint • The validity of the estimation method is discussed based on a result of experiment.

We1-6(5) 11:20-11:35 We1-6(6) 11:35-11:50 Design for Linkage Mechanism Based on Development of a Rotary-percussive Drilling Reliability of Kinematic Accuracy Mechanism (RPDM)

Nianfeng Wang, Shuai Wei and Xianmin Zhang Qiquan Quan, Peng Li, Shengyuan Jiang, Xuyan Hou, Dewei Guangdong Province Key Laboratory of Precision Equipment and Tang, Zongquan Deng, Weiwei Zhang Manufacturing Technology, South China University of Technology, China School of Mechatronics Engineering, Harbin Institute of Technology , China

• The reliability analysis has been applied in the • RPDM have both rotation and percussion optimization design about the four-bar mechanism functions, and it can drive the coring drill work of a mechanical finger. at either rotary or rotary-percussive mode. • With the actual requirements of the finger, the • A cam driving mechanism is adopted in RPDM optimization of the mechanism has been carried out to generate impact for improving hard rock by genetic algorithm, and the optimization results breaking capability. are certain rod lengths and the initial pose of the four-bar mechanism. The four-bar mechanism with • The mechanic analysis and dynamic simulation reliability design has a good motional smoothness of the driving mechanism are implemented. Mechanical design of a which is not easy to fail. finger Rotary-percussive Drilling Mechanism (RPDM)

54 IEEE ROBIO 2012 Conference Digest

We1-7: Service Robots II Session Chairs: Toshikatsu Suzuki and Chi Zhu

Room: Jade 7, 10:20-12:05, Wednesday, December 12, 2012

We1-7(1) 10:20-10:35 We1-7(2) 10:35-10:50

The Control System Design of Automatic Weeding A SIFT-Based Person Identification Robot Based on Visual Navigation using a Distance-Dependent Appearance Model Chuanbo QIn, Qiliang Du, Lianfang Tian for a Person Following Robot Automation Science and Engineering, South China University of Technology, China Junji Satake, Masaya Chiba, and Jun Miura Xiaogang Huang Department of Computer Science and Engineering, Mechanical and Automotive Engineering, Toyohashi University of Technology, Japan South China University of Technology, China • We propose a person identification target person • Realize robot automatic change line in fields . technique for a mobile robot which other people with similar clothing • Visual navigation based on DM6446 and Linux. follows a specific person. • To cut off weeds with the crawler mobile • The method is based on SIFT features mechanism in stamping model. of the clothing. • motion control board was designed based on • We make appearance models for DSP,FPGA for movement control and feedback various body directions, and set the signal sample. distance-dependent threshold. • Experimental results show that the robot Following a specific person by our mobile robot. Weeding robot successfully followed a specific person.

We1-7(3) 10:50-11:05 We1-7(4) 11:05-11:20

Posture Estimation using a 3D Range Camera Running on a Slope with Power Assistance for an for Standing-Up Motion Assist Robot Omnidirectional Hybrid Walker and Wheelchair Toshikatsu Suzuki*, Chi Zhu*, Masataka Yoshioka*, Shota Simazu*, Yuichiro Yosikawa*, Yuji Okada*, Yuling Yan**, Haoyong Yu***, Natsumi Tsukada, Kazuya Ogura, Aiguo Ming and Makoto Shimojo and Feng Dung**** Dept. of Mechanical Eng. and Intelligent Systems * Maebashi Inst. of Tech., Japan. ** Santa Clara Univ., USA. The Univ. of Electro-Communications, Japan *** National Univ. of Singapore, Singapore. **** Nankai Univ., China. • A novel kind of omnidirectional hybrid walker and wheelchair • To realize safe assist by real time feedback is developed and the movement of the robot traveling control based on the information from a high throughout a slope is analyzed. speed tactile sensor and a 3D range camera • The intrinsic fact is firstly clarified that the speed difference is • This paper describes the posture estimation of inevitable between the user and the user-propelled walker or human using a 3D range camera. wheelchair when starting to ascend/descend a slope. • A method for posture estimation from the 3D • The above result explains why the user’s load suddenly increases/decreases when ascending/descending a slope range data is proposed and implemented. system for Experimental results show the availability of human assist • Speed compensation approach is proposed to get smooth the 3D range camera for posture estimation. "load feeling" as traveling on horizontal plane and verified by experimental results.

We1-7(5) 11:20-11:35 We1-7(6) 11:35-11:50

Gradient Calibration for the RCBHT Cantilever Study on cooperation between humanoid Snap Verification System robot Nao and the Barrett WAM J. Rojas, K. Harada, H. Onda, N. Yamanobe, E. Yoshida, K. Nagata, and Y. Kawai Tingting Liu AIST School of Control Science and Engineering, Shandong University, China • Goal: Max Q.-H. Meng • To generalize the use of the RCBHT system in identifying the robot Department of Electronic Engineering, The Chinese University of Hong Kong, state of cantilever snap assemblies (“Snap Sensing”). China • Method: • By exploring key contact • Nao communicates with the user and regions under the RCBHT obtains user’s instructions, then sends system, gradient markers were identified. them to WAM. • Gradient schemes were identified to generalize the • WAM gets the commands and executes parameterization of RCBHT primitives at the lower level. the relevant tasks. • Foundation: • Three methods using three mediums to • Constrained assemblies and command Nao are studied and modified. System architecture extrapolated intuitive context in signal prediction allow to generalize parameter optimization.

55 IEEE ROBIO 2012 Conference Digest

We2-1: Micro/nano Devices & Robots Session Chairs: Markus Kagerer and Kundong Wang

Room: Jade 1, 13:30-15:30, Wednesday, December 12, 2012

We2-1(1) 13:30-13:45 We2-1(2) 13:45-14:00

Manufacturing of patient-specific pancreas Design of an Aerostatic Imprint Head with Off-plane models for surgical resections Alignment Functionality for Nanoimprint Lithography J. Schwaiger, M. Kagerer, M. Träger and T. C. Lueth Chong Du, Weihai Chen, Yunjie Wu, and Mei Yuan Institute of Micro Technology and Medical Device Technology, Technische Dep. Automation Science and Electrical Engineering, Beihang University, China Universität München, Germany Wenjie Chen S. Gillen, M. Dobritz, J. Kleeff and H. Feussner Singapore Institute of Manufacturing Technology, Singapore Department of Surgery / Radiology, Technische Universität München, Germany • Anatomic models are applied in medical • Spherical aerostatic bearing allows to make off- education and surgical training. A figure here plane alignment with extremely low friction. • Segmentation of pancreas / pancreatic cancer • Compliant structure for high precision alignment and surrounding vessel structures. is out of the force loop. • Rapid prototyping of patient-specific, real 3D • Off-plane alignment does not affect the in-plane models. alignment. • Manufacturing of pancreas models is • Stiffness in z direction is improved with magnetic promising for support of planning of pancreatic preload structure. A photograph of the cancer surgery. developed device

We2-1(3) 14:00-14:15 We2-1(4) 14:15-14:30

Design and Analysis of a New Flexure-based XY Micro- positioning Stage with Decoupled Motion Characteristic Xiantao Sun, Weihai Chen, Rui Zhou and Jianbin Zhang School of Automation Science and Electrical Engineering, Beihang University, China Inchworm-like Micro Robot for Wenjie Chen Human Intestine Singapore Institute of Manufacturing Technology, Singapore • The proposed XY flexure stage has kinematic decoupling characteristic in a Kundong Wang and Xuan Jin small deflection. Shanghai Jiao Tong University, China • The analytical model of the coupling error is derived. • The dynamic model of the manipulator is established using an equivalent pseudo-rigid-body method. High-precision XY flexure stage

We2-1(5) 14:30-14:45 We2-1(6) 14:45-15:00

An Inchworm Robot using Electro-conjugate Fluid

Shouhei Ueno, Kenjiro Takemura, Shinichi Yokota, and Kazuya Edamura Milling Force Mixed-signal Denoising Based on Micro-nano Mechatronics lab., Keio University, Yokohama, Japan ICA in High Speed Micro-milling

Yue Sun, Yongsheng Liu, Zhanjiang Yu, Huadong Yu, Jingkai Xu • An electro-conjugate fluid (ECF) is a kind of functional fluid, which produces Abstract a powerful flow (ECF jet) In order to obtain the real milling force signal in high speed micro- milling process, and accurately identify each exciting source, this paper studies • We pay our attention to the motion on micro-milling force mixed-signal separation and identification technology based pattern of inchworm because an on combination of independent component analysis (ICA) and fast Fourier transform (FFT). The ICA ethod from theory of blind source separation combining inchworm is highly robust against the is used with FFT to separate and identify the micro-milling force mixed- moving environment signal collected from the dynamometer, and the independent micro-milling force signal and noise signals are extracted. ICA theory shows this method is suitable to • We developed the inchworm robot Sucker separate both Gaussian signals and non-Gaussian signals. ICA could make up having suckers which is driven by the for the shortcomings of traditional methods which can only inhibit Gaussian noise ECF The inchworm robot signals. Using this method, the experiments successfully separate micro- milling force signal, non-Gaussian machining noise signal and Gaussian environmental noise signal.

56 IEEE ROBIO 2012 Conference Digest

We2-2: Biomimicking III Session Chairs: Gi-Hun Yang and Minoru Hashimoto

Room: Jade 2, 13:30-15:30, Wednesday, December 12, 2012

We2-2(1) 13:30-13:45 We2-2(2) 13:45-14:00 Numerical Simulation of Odor Plume in Indoor Finding and Utilizing Self-preserving Functions in Blood Pulsation Ventilated Environments for Studying Odor Source -Development for Ethanol Based Artificial Circulation System- Localization with Mobile Robots Xiao-Bao Jiang, Qing-Hao Meng, Yang Wang, Ming Zeng, Chyon Hae Kim and Wei Li Honda Research Institute Japan Co. Ltd., Japan School of Electrical Engineering and Automation, Tianjin University ,China Yuki Nishi, Syunsuke Nagahama and Shigeki Sugano Waseda Universtiy, Japan • To facilitate the research of mobile robot based odor source localization (OSL). ・How the circulation systems of Pure Fuel • The wind field is obtained by superimposing a humans and animals are supporting Filter Mixed Fuel Fuel Pure Fuel normal-distribution fluctuating speed on the their survival capability have not been Waste Fluid Mixed Fuel calculated advection velocity. studied sufficiently. Filtering System • The odor plume is built by combining the Farrell’s ・In this paper, we tried to reveal the Flow Meter Fuel Tank filament concentration model with the simulated relationship between heart pulsation wind field. and blood clotting using a biomimetic Test Point Pump circulation system. • Through comparing the wind and concentration Fig. 1. Indoor information , the creditability of the established ・We got strong evidence that blood simulation plume Diagram of Biomimetic Circulation System simulated plume is demonstrated. clotting is supported by heart beating.

We2-2(3) 14:00-14:15 We2-2(4) 14:15-14:30

Obstacle Avoidance Research of Snake-like Implementation and Navigation of a 3-DOF Fish Robot Based on Multi-sensor Information Robot ‘ICHTHUS V5.5’ Gi-Hun Yang, Sang-Hyo Lee, Hyun-Jin Lee, and Young-Sun Ryuh Qianying Wu, Junyao Gao and Chengzu Huang Biomimetic Robot Research Group, KITECH, Korea Intelligent Robotics Institute, Beijing Institute of Technology, China • A robotic fish ‘ICHTHUS V5.5’ is introduced • This paper is studied based on the rope-drive with its main components in the point of design and implementation snake-like robot. A figure here A figure here • Information fusion method is adopted to • The mathematical model of the robotic fish integrate the collected information. was derived for which swims using tail fin of 3-DOF link mechanism • Laser radar and ultrasonic sensors are used to detect the surrounding environment. • Localization and point-based autonomous navigation is successfully achieved in the • The snake-like robot can successfully avoid Actual Implementation of Robotic real environment through those obstacles after training of the fuzzy neural The rope-drive snake-like Fish 'Ichthus V5.5’ robot experimental results network.

We2-2(5) 14:30-14:45 We2-2(6) 14:45-15:00

Muscle Roles on Directional Change during Temporal Gait Parameters Captured by Hopping of a Biomimetic Feline Hindlimb Surface Electromyography Measurement Andre Rosendo, Kenichi Narioka and Koh Hosoda Chi Chong Chan and Wei-Hsin Liao Adaptive Robotics Laboratory, Osaka University, Japan Department of Mechanical and Automation Engineering, The Chinese University of Hong Kong, Hong Kong, China • Design and development of a quadruped robot. • Biomimetic hindlimb approach. • The low frequency of SEMG signals (< 20 Hz) can capture the motion’s • Exploration of biomimetic hindlimb characteristics. motion range with vertical hopping. • Linear envelope, linear prediction, • Possible biological explanation for and threshold analysis are applied to directional changes on feline jumps. process the SEMG signals. • Temporal gait parameters, swing period, stance period and stride Output of linear prediction period of walking can be obtained. process and muscle activities

57 IEEE ROBIO 2012 Conference Digest

We2-3: Medical & Surgery Robots III Session Chairs: Xiaorui Zhu and Kazuhito Yokoi

Room: Jade 3, 13:30-15:30, Wednesday, December 12, 2012

We2-3(1) 13:30-13:45 We2-3(2) 13:45-14:00

A Laser Sintered Miniature Gear Hub for a Workflow Analysis and Surgical Phase Surgical Robot and its Maximum Torque Recognition in Minimally Invasive Surgery

Mattias F. Traeger, Daniel B. Roppenecker and Tim C. Lueth Surgical phases of a single port sigma resection are detected with a Institute of Micro Technology and Medical Device Technology (MIMED), Input-Output HMM with a Bayes classifier Technische Universitaet Muenchen, Germany

A figure here

Laser sintered miniature Single port robotic setup gear hub for a servomotor

1 29.11.2012 1 Institute for Process Control and Robotics (IPR)

We2-3(3) 14:00-14:15 We2-3(4) 14:15-14:30 Improvement of Dynamic Transparency of Design of a Miniaturized Active Endoscopic Haptic Devices by Using Spring Balance Capsule with Revolving Head Mechanism

Rongfang Fan, Chen Zhao and Hanwang Zhao Xiaorui Zhu, Chunxin Qiu, Zhidong Shi the Key Laboratory for Mechanism Theory and Equipment Design of Ministry of Harbin Institute of Technology Shenzhen Graduate School, Shenzhen, China Education, Tianjin University, China • An active endoscopic capsule with a simplified construction • Force-reflecting devices should be static and revolving head mechanism is balanced to get good transparency. designed. • This paper presents a mixed approach which • The finite element method is used uses counterweights to balance the orientation to analyze and optimize the mechanism and springs to the position electromagnetic driven head- mechanism of the haptic interface. swinging module on the capsule. Experiments were carried out to • As an example a PHANTOM haptic device is • demonstrate the proposed design The experimental validation of the applied. The results show that the device could The Effect of Gravity Balancing is effective to achieve the desired revolving head mechanism be balanced in a large enough space with only performance. a small amount of imbalance left.

We2-3(5) 14:30-14:45 We2-3(6) 14:45-15:00

Image-Guided Steering of a Motorized Hand-Held Quantitative Evaluation of Human Rising Flexible Rhino Endoscope in ENT Diagnoses Movements from Chair in Terms of Physical Strain Imposed on Joints Chen Fang, Weiwei Sang, Jan D.J. Gumprecht and Tim C. Lueth Kazunori Wada, Keisuke Hagiwara, Toshikazu Matsui, and Institute of Micro- and Medical Device Technology, TU Munich, Germany Nobuaki Nakazawa Gero Strauss Graduate School of Engineering, Gunma University, Japan Department of ENT, University Hospital of Leipzig, Germany

(1) t1=100 msec, ω14=0.041 (2) t1=200 msec, ω14=0.197

• Objective: to quantitatively evaluate human (3) t1=300 msec, ω14=0.541 (4) t1=400 msec, ω14=1.2 =500 msec, ω =2.8 (6) t =600 msec, ω =6.64 • Image based steering of a flexible (5) t1 14 1 14 sit-to-stand movements in terms of physical (7) t =700 msec, ω =19.55 rhinoscope for ENT diagnosis 1 14 strain imposed on the human joints. knee (1)  (7) • A hand-held lightweight manipulator as • Method: to derive joint-torque profiles during

a adapter to motorize the endoscope m] sit-to-stand movements using the dynamics- ･ • Image processing to find the ROI switching optimal control model.

• Semi-automatical control of the [N Torque • Result: there exists three optimal conditions ankle (1)(7) flexible rhinoscope to tract the ROI for the joint torques to take respective minima. hip (1)  (7)

• Conclusion: the human sit-to-stand mecha- Time [sec] nism minimizes the physical strain on joints. Fig. 7 Model’s joint torque profiles

58 IEEE ROBIO 2012 Conference Digest

We2-4: Human Robot Interaction I Session Chairs: Yi Zhang and Jian Huang

Room: Jade 8, 13:30-15:30, Wednesday, December 12, 2012

We2-4(1) 13:30-13:45 We2-4(2) 13:45-14:00

A Multi-Sensor Armband based on Muscle and Hand Tracking and Pose Recognition via Depth Motion Measurements and Color Information Cheng Tang, Yongsheng Ou, Guolai Jiang, Qunqun Xie James Cannan and Huosheng Hu Shenzhen Institutes of Advanced Technology, Chinese Academy of Sciences, School of Computer Science and Electronic Engineering, China University of Essex, Colchester CO4 3SQ, UK Yangsheng Xu  Combining motion sensors such as The Chinese University of Hong Kong, Hong Kong SAR, China gyro, accelerometer and • In this paper, we have tackled the problems of magnetometer, with an EMG muscle A figure here hand tracking and pose estimation by sensor. analyzing real-time depth data and RGB data  Comparison between Fusion Band obtained from the Kinect sensor. and commonly used industrial • It is shown in the experiments that our system control techniques. can track the hand robustly.  With improvements, real world • And recognize more than 90% of the hand poses we define for our depth image Example for Hand Tracking commercial applications would be and Pose Recognition possible. Fusion Band and database. robotic arm.

We2-4(3) 14:00-14:15 We2-4(4) 14:15-14:30

A Trifocal Tensor Based Camera-Projector Tool Centered Learning from Demonstration for System for Robot-Human Interaction Robotic Arms with Visual Feedback

Jinglin Shen and Nicholas Gans A. S. Phung, J. Malzahn, F. Hoffmann and T. Bertram Department of Electrical Engineering, the University of Texas at Dallas, USA Institute for Control Theory and Systems Engineering, TU Dortmund, Germany

• Reproduction and generalization of • We present a trifocal tensor based movement from demonstrations based camera/projector system for robot-human on visual feedback interaction • Description of movement pattern using • Saliency based object detection dynamic movement primitive (DMP) • Object matching in two camera views based system on shape comparison and trifocal constraint • Learning straight point-to-point and more • Projector provides feedback information on complex trajectories with respect to the detected objects in the workspace object pose • A robot manipulator is guided by the vision • Experimental results on Katana arm system to approach and grasp an object

We2-4(5) 14:30-14:45 We2-4(6) 14:45-15:00

Measurement of the Relationship between Analyzing Stability of Haptic Interface using Weight Perception and Upper Limb Movement Linear Matrix Inequality Approach Quoc Viet Dang, Laurent Vermeiren, Jian HUANG, Tetsuya TANINO, Kazuki WADA, Antoine Dequidt and Michel Dambrine Nobuyasu TOMOKUNI LAMIH-UMR CNRS 8201, University of Valenciennes School of Engineering, Kinki University and Hainaut-Cambrésis, France • Mechanical model of haptic interface. • Investigation of the relationship between • Discrete-time state-space representation motion of the human upper limb and of the closed-loop haptic system: weight perception during weight lifting experiments in a real environment. • Psychometric and psychophysical magnitude functions were computed. • Stability condition in term of Linear Matrix Inequality: • The slope of the psychometric function and the modality characteristic exponent • Algorithm for reconstructing the stability were quantitatively analyzed region in virtual damping-stiffness (Bve- Kve) plan. Force feedback haptic system

59 IEEE ROBIO 2012 Conference Digest

We2-5: Aerial Robots II Session Chairs: Qiang Zhan and Ertugrul Cetinsoy

Room: Jade 5, 13:30-15:30, Wednesday, December 12, 2012

We2-5(1) 13:30-13:45 We2-5(2) 13:45-14:00

3D PRM based Real-time Path Planning for UAV An Implement of RPV Control System for Small in Complex Environment Unmanned Helicopters Fei Yan and Jizhong Xiao Department of Electrical Engineering, The City College of New York, USA Yicheng Zhang and Tianmiao Wang, Jianhong Liang, Chaolei Yan Zhuang Wang,Yang Chen, Yi Zhou, Yubao Luan,Han Gao Faculty of Electronic Information and Electrical Engineering, Dalian University Robotic Institute Beihang University, China of Technology, China • The work space is represented by free voxels, • Develop an EKF-based GPS/INS NAV which are big enough for UAV navigation. A figure here system with low-cost sensors. A figure here • Evaluation of voxel connectivity is more • A chaos-genetic Alg. is proposed in the efficient in bounding boxes than in the whole system identification of the Module. space. • A LQG-based RPV (Remotely Piloted • Improved PRM is proposed to ensure more Vehicle) control scheme is applied in even distribution of the nodes in work space. the control system. • Finally, A* algorithm is carried out to search • The flight of the control system has For figure caption (optional), path in roadmap. been verified by simulation use arial narrow 20pt font environment and real experiment.

We2-5(3) 14:00-14:15 We2-5(4) 14:15-14:30

Aerial Localization of an RF Source in NLOS Condition Control System Design and Experiments of a Quadrotor S.M.M. Dehghan, M. Farmani, H. Moradi, Qiang Zhan, Junqing Wang and Xi Xi school of Electrical and Computer Engineering, University of Tehran, Iran Robotics Institute, Beihang University, China

• Localization of an RF source such as a • An ARM based control system with low-cost cell phone in a search and rescue A figure here and low-weight inertial sensors were designed. mission; • The modeling, attitude estimation and attitude • Unknown power and None Line Of control of the quadrotor were presented. Sight (NLOS) condition; • Experimental validation of the control system Two new approaches to benefit from • A UAV hovers over a was presented. both RSSI and AOA based localization Flying experiments of a terrain searching for a RF quadrotor methods; signal to localize the source

We2-5(5) 14:30-14:45 We2-5(6) 14:45-15:00

Design and Simulation of a Holonomic Dynamic IBVS Control of an Underactuated UAV Quadrotor UAV with Sub-Rotor Control Surfaces Hamed Jabbari and Hossein Bolandi Ertugrul Cetinsoy Electrical Engineering, Iran Univ. Science & Tech., Iran Faculty of Engineering and Design, Istanbul Commerce University, Turkey Giuseppe Oriolo Ingegneria Informatica, Università di Roma La Sapienza, Italy

• Dynamics of the quadrotor helicopter is • Quadrotor helicopter UAV with sub-rotor considered in designing an image based control control surfaces for non-inclined horizontal A figure here strategy. motion • Inertial information of the robot is used to • Dynamic model of the air vehicle provide passivity properties for dynamics of • Eight output PID control system for controlling perspective image moments. both the rotor speeds and control surface • selected image moments generates satisfactory angles trajectories in Cartesian coordinates. Dynamic IBVS in virtual • Simulation results with severe wind conditions • A nonlinear controller for the full dynamics of image plane the system is designed.

60 IEEE ROBIO 2012 Conference Digest

We2-6: Smart Structures, Materials, Actuators III Session Chairs: Haoyong Yu and Mahdy Eslamy

Room: Jade 9, 13:30-15:30, Wednesday, December 12, 2012

We2-6(1) 13:30-13:45 We2-6(2) 13:45-14:00 Research on the Relationship between the Oscillating Vibrator Parameter and the Working Ability of Caudal- Kinematic Analysis and Optimal Design of a fin-type Valveless Piezoelectric Pump 3D Compliant Probe Liang Wang, Jianhui Zhang and Jiamei Jin State Key Laboratory of Mechanics and Control of Mechanical Structures, Zhang Xuchong, Chen Zhong, Zhang Xianmin Nanjing University of Aeronautics and Astronautics, China Key Laboratory of Precision Equipment and Manufacturing Xiaoqi Hu Technology of Guangdong Province Institute of Technology, Lishui University, China South China University of Technology, China. • Expounding the structure design and elaborating the operating principle of oscillating vibrator.  A 3D compliant probe taking the • The theoretical model of CVPP is established and advantage of delta structure and theoretical flow rate is analyzed. flexure hinges.  • The relationships between oscillating vibrator Kinematic analysis of the probe parameter and the working ability of pump are using screw theory and stiffness conducted. matrix method  The optimized model of the 3D • The working ability of CVPP was enhanced by compliant probe is obtained. optimizing the parameters of the oscillating The relationships between pressure Prototype of 3D probe vibrator. head and the length of VLOV

We2-6(3) 14:00-14:15 We2-6(4) 14:15-14:30

An attempt to make a large-scale stacked-type Design and Analysis of a Novel Compact electrostatic actuator for artificial muscles of robots Compliant Actuator with Variable Impedance Makoto Ito Haoyong Yu, Sunan Huang, Gong Chen, Siew-Lok Toh, Manolo STA Cruz Department of Physics, Tokyo Institute of Technology, Japan National University of Singapore, Singapore Keiji Saneyoshi Yessine Ghorbel, University of Stuttgart, Germany Radiation Research and Management Center, Chi Zhu, Maebashi Institute of Technology, Japan Tokyo Institute of Technology, Japan Yuehong Yin, Shanghai Jiaotong University, China • An electrostatic actuator which is scaled larger • We proposed novel series elastic actuator easily to get larger force and longer stroke. (SEA)design that overcomes the common • The structure prevents the air gap between the limitations of current SEAs electrodes from unnecessarily widening. • The compact actuator design is suitable for • The prototype achieved about 150mN at 740V wearable robotics for rehabilitation at the natural length. • Detail design and modeling of the actuator is Physical Prototype • The properties will improve by decreasing the presented hinges’ thickness and increasing the adhesive • Experimental results on force control have strength. validated the design

We2-6(5) 14:30-14:45 An Easily Manufactured Structure and Its Analytic Solutions for Forward and Inverse Position of 1-2-3-SPS Type 6-DOF Basic Parallel Mechanism Tongzhu Yu, Huiping Shen*, Jiaming Deng et.al Institute for Robot, Changzhou University, China

• Propose an easily manufactured structure for 1-2-3-SPS type 6-DOF basic parallel mechanism. • Propose a method of solving forward position based on the analysis of coupling degree. • The forward position solutions are verified by performing an inverse position analysis .

1-2-3-SPS type 6-DOF basic parallel mechanism

61 IEEE ROBIO 2012 Conference Digest

We2-7: Service Robots III Session Chairs: Ryo Hanai and Yohan Noh

Room: Jade 7, 13:30-15:30, Wednesday, December 12, 2012

We2-7(1) 13:30-13:45 We2-7(2) 13:45-14:00

Study of Architecture Based on Doorway Passing of an Intelligent Wheelchair by Middleware and Abstract Environment Dynamically Generating Bézier Curve Trajectory Qiongxiong Ma, Yanbiao Zou and Tie Zhang School of Mechanical & Automotive Engineering , South China University of S. Wang, L. Chen, H. Hu and K. McDonald-Maier Technology , China School of Computer Science and Electronic Engineering, University of Essex, Colchester CO4 3SQ, UK

• Described the run principle and the members • Novel door passing strategy for wheelchair to of the service architecture . pass a doorway using local coordinates. • Designed the abstract environment and the • Dynamic generation of Bézier curve trajectory mechanism of the environment change event in real time. • Extended the event mechanism to suit for the • Laser based real-time door finding algorithm. different platform via the communication platform • Experimental results show the good performance on finding the door and passing • Described the robot application develop the doorway. Door Passing framework for improving software reusability Experimental Scenario and expediting robot development Architecture and Implementation

We2-7(3) 14:00-14:15 We2-7(4) 14:15-14:30

Development of an assistive device for elderly Proposal of Architecture and Implementation Process people: Fuzzy Sensor Fusion Experimental Study for IEC61508 Compliant, Dependable Robot Systems Omar Salah, Ahmed A. Ramadan and Ahmed A. Abo-Ismail Ryo Hanai, Yoshihiro Nakabo, Kiyoshi Fujiwara,Takuya Ogure, Mechatronics and Robotics Engineering Department, EJUST, Egypt Daichi Mizuguchi, Keiko Homma and Kohtaro Ohba Makasatsu Fujie, and Atsuo Takanishi Intelligent Systems Research Institute, AIST, Japan Graduate school of creative science and technology, Waseda University , Japan Hajime Saito General Robotix, Inc., Japan • This paper describes a prototype of an assistive device for elderly people . • Dependability & accountability is becoming important for robots. • Several experiments were carried out using VICON system and inertial sensors to determine • Architecture & development process the human posture while sit to stand motion. considering IEC61508, which is a standard for functional safety is • ANFIS algorithm is applied to convert the presented. inertial sensors signals to the right posture of the patient. • The main target is systems of robots EJAD that work closely with humans.

We2-7(5) 14:30-14:45 We2-7(6) 14:45-15:00 Development of the Airway Management Training Study on a Practical Robotic Follower System WKA-5: Improvement of Mechanical Designs to Support Home Oxygen Therapy Patients for High-Fidelity Patient Simulation - Prototype Cart Development Applying the Inverted Pendulum Control - Yohan Noh, Chunbao Wang, Mitsuhiro Tokumoto, Yusuke Masatsugu Iribe and Ryoichi Dasai ( O.E.C.U. , Japan) Matsuoka, Terunaga Chihara, Hiroyuki Ishii, Atsuo Takanishi Gen Endo ( T.I.T. , Japan ) Toshio Takubo ( T.W.U., Japan ) Graduate School of Science and Engineering, Waseda University, Tokyo, Japan Toshiyuki Takayama, and Satoru Shoji Tetsuya Kinugasa ( O.U.S., Japan ) Koichi Osuka ( Osaka Univ., Japan ) Kyoto Kagaku Co. Ltd., Kyoto, Japan • Present mechanical mechanism designs which • Home Oxygen Therapy patients need to take consider internal organs and external organs the high level oxygen tank with cart. anatomically for airway management. • Mass of the tank and cart is heavy enough to • Improve the mandible mechanism and the discourage the patients from going out. tongue mechanism better than previous system • We developed new prototype robotic cart for • A set of experiments was carried out to doctor carrying oxygen tank of the patients subjects in order to verify the usefulness of our • In this paper we describe the effectiveness of proposed mechanism, and discuss the doctors the prototype robotic cart. on the results of the experiments. Mechanical mechanism of WKA-5

62 IEEE ROBIO 2012 Conference Digest

We3-1: Intelligent Systems I Session Chairs: Ronald Arkin and Markus Kagerer

Room: Jade 1, 15:20-16:50, Wednesday, December 12, 2012

We3-1(1) 15:20-15:35 We3-1(2) 15:35-15:50

Primate-inspired Mental Rotations: Piezo Inkjet Drop-on-Demand Experimentation Implications for Robot Control Platform Manufactured with Rapid Prototyping Ronald C. Arkin, Frank Dellaert, and Joan Devassy Techniques Enabling Future Technologies School of Interactive Computing Markus Kagerer, , Kenji L. Eiler, Thomas Georgia Institute of Technology, USA Student Member, IEEE Ottnad, Franz Irlinger, and Tim. C. Lueth, Member, IEEE Institute of Micro Technology and Medical Device Technology, • Examines the cognitive process of mental Technische Universität München, Germany rotation in primates as inspiration for robot navigational control. • Experimentation platform based on an inkjet • Assesses the role of optic flow in support of printhead and on a support plate representations used in mental rotations. • Describes envisioning as a method for • Inkjet printhead is manufactured in less than integrating this cognitive computation. 30 minutes due to rapid prototyping techniques • Includes preliminary results using depth data • Liquid media and media with solid-liquid phase derived from optic flow and depth sensors. transition are ejectable Experimentation platform

We3-1(3) 15:50-16:05 We3-1(4) 16:05-16:20

Map Integration of Human Trajectory with Sit- An Online Gait Adaptation with SuperBot in ting/Standing Position using LRF and Kinect Sensor Sloped Terrains Teawon Han, Nadeesha Ranasinghe, Luenin Barrios, Yuma NIHEI1,2,3, Ippei SAMEJIMA1,2,3, Naotaka HATAO1,3 and Wei-Min Shen Hiroshi TAKEMURA1,2, Satoshi KAGAMI1,2,3 Information Sciences Institute, University of Southern California, USA 1: The National Institute of Advanced Industrial Science and Technology, Japan Memory 2: Tokyo University of Science, Japan • Intelligent Online Reconfiguration System in every step 3: Japan Science and Technology Agency, Japan • Recognizing feature of environment (slope) learning - Time per a step, Accelerometer sensor What is reasoning • Detecting human sitting/standing posture and - Sensed data Distribution (ds) the best human trajectory. - Likelihood is updated (L) by MEC gait? Superbot Module • Integrating them as map information. - Distribution (ds) Belief of slope (db) Choosing Recognizing • Overcoming the weak point of Kinect by using • Learning Chosen gait’s preference feedback LRF and mobile robot. - Time per a step • Choosing the best gait among given gaits Sloped Gaits -∑(belief of slope (db) × preference of each gait) Terrains

We3-1(5) 16:20-16:35 We3-1(6) 16:35-16:50

Dissolved Oxygen Control of Wastewater Bionic Eye System Based on Fuzzy Adaptive PID Treatment Using Composite Fuzzy Control Control

Wenhao Shen, Erpan Tao, Li Ning and Tianlong Liu State Key Lab of Pulp & Paper Eng., South China University of Technology YuMei Zhou, Jun Luo, JuQi Hu, HengYu Li, ShaoRong Xie Mechatronics Engineering, Shanghai University, China • Control DO concentration in a pilot SBR of papermaking wastewater treatment • The mechanism can realize 3-DOF motion namely process; Pan/Tilt/Zoom like human eyes. A figure here • Four control strategies are studied: • Fuzzy Adaptive PID Control method is applied as ON/OFF, PID, fuzzy and composite fuzzy the equivalent of vestibular nuclei ’s function. (CF); • The main control unit is based on • CF obtains good controlling performance DSP/TMS320F2812 processor. at the price of relatively high aeration Spherical Parallel Mechanism volume. Schematic of the SBR pilot process

63 IEEE ROBIO 2012 Conference Digest

We3-2: Biologically Inspired Robotics III Session Chairs: Kohei Nakajima and Minoru Hashimoto

Room: Jade 2, 15:20-16:50, Wednesday, December 12, 2012

We3-2(1) 15:20-15:35 We3-2(2) 15:35-15:50

Active Walking Robot Mimicking Local Information Transfer in Soft Robotic Arm Flat-Footed Passive Dynamic Walking K. Nakajima1, T. Li1, R. Kang2, E. Guglielmino2, D. G. Caldwell2, and R. Pfeifer1 Yuta Hanazawa, Hiroyuki Suda, Yu Iemura and Masaki Yamakita 1Department of Informatics, University of Zurich, Switzerland Mechanical and Control Engineering, Tokyo Institute of Technology, JAPAN. 2Department of Advanced Robotics, Istituto Italiano di Tecnologia, Italy

(i)Interaction with the environment • Characterizing the diverse body dynamics of • We present a novel active walking robot soft robots. that can mimics flat-footed passive • Permutation-information theoretic approach for dynamic walking (FFPDW). soft robots. (ii)Spatiotemporal dynamics • Our flat-footed biped robot has springs 21 • Propose a method for visualizing the and inerters at each ankle. points information transfer spatiotemporally through 1 • We show that our robot achieves fast (iii)Local information transfer the robot’s body. 21 active walking similar to FFPDW on level ground. • Demonstrating the power of the measures by points 1 4500 4520 4540 4560 4580 4600 using the soft robotic arm simulator and timestep physical platform inspired by octopus. Our active biped robot

We3-2(3) 15:50-16:05 We3-2(4) 16:05-16:20

Adaptive RBFNN Type-2 Fuzzy Sliding Mode A Method for Optimal Design of an Inchworm Controller for Robot Arm with Pneumatic Muscles Climbing Robot Rezoug Amar and Hamerlain Mustapha Robotic laboratoiry, Centrer for Developement of Advanced Technologies, Algiers, Algeria David Pagano, Dikai Liu and Kenneth Waldron Tadjine Mohamed Centre for Autonomous Systems, University of Technology, Sydney, Australia Control Process Laboratoiry, Polytechnic National Schcool of Algiers,Algiers Algeria • Digest must be prepared in MS Office • There are a number of design challenges when PowerPoint and we accept no other format. A figure here designing an inchworm for specific applications • Use Arial 28pt font in bold face for the title. • Environmental features, robot mobility and • Use Arial 24pt font for the authors and Arial safety factors must all be considered 20pt font their brief affiliations. • A design model has been developed to • Use 3 or 4 bullet points (limit each to less than optimise the configuration to meet these 15 words in Arial Narrow 24pt font), and only 1 design challenges figure (replace the figure on the right with your For figure caption (optional), • Optimisation results show increased figure). use arial narrow 20pt font configuration mobility and robustness within the application

We3-2(5) 16:20-16:35 We3-2(6) 16:35-16:50

A Simplified CPGs Network with Phase oscillator Modeling and Experimental Investigation of Model for Locomotion Control of Snake-like Robot Viscoelastic Contact between a Spiral-type Norzalilah Mohamad Nor and Shugen Ma Robotic Capsule and a Gastrointestinal Tract Department of Robotics, Ritsumeikan University, Japan Hao Zhou, Gursel Alici, Trung Duc Than and Weihua Li Faculty of Engineering, University of Wollongong, Australia • A novel structure of CPG network is proposed to control locomotion of a snake-like robot. • FEA is used to predict viscoelastic contact • Two main points addressed: 1) simple network between a spiral-type capsule and a structure with less mathematical computation, gastrointestinal tract. and 2) a single parameter to control the body shape of the snake-like robot. • The prediction is reasonably consistent with the experimental result measured by a custom- • Simulation results show that the proposed built setup. CPG network can be used effectively to control Our proposed CPG network the serpentine locomotion of a snake-like implemented into snake-like robot • It shows the feasibility of the FEA as a tool to robot. evaluate the viscoelastic contact. • It helps to design and optimize the traction topology of a spiral-type robotic capsule.

64 IEEE ROBIO 2012 Conference Digest

We3-3: Sensor Networks Session Chairs: Hongliang Ren and Tiantian Feng

Room: Jade 3, 15:20-16:50, Wednesday, December 12, 2012

We3-3(1) 15:20-15:35 We3-3(2) 15:35-15:50

A fully-autonomous hovercraft inspired by bees: wall Behavior-based Fuzzy Control Framework for following and speed control in straight and tapered corridors Multi-Mobile-Robot Deployment

F. L. Roubieu, J. Serres, N. Franceschini, F. Ruffier and S. Viollet Xu Zhong and Yu Zhou Aix-Marseille University, CNRS, ISM UMR 7287, Marseille, France Mechanical Engineering, State University of New York at Stony Brook, USA

• The robot (A) embeds a 8.17-gram and 0.75W- power consuming visuo-motor control system • A novel distributed motion control framework inspired by bees, for deploying multiple mobile robots • The autopilot consists of two lateral optic flow • Accomplished using a behavior-based fuzzy regulators, which jointly control speed and control architecture clearance to the walls, • Resulting in effective sensory coverage, • The hovercraft equipped with this optic-flow- communication connections, and collision-free based autopilot navigates along a tapered (B) (A) The LORA robot. target approaching or straight (C) corridor up to 0.8 m/s. Chronophotography of the robot travelling in a tapered (B) and in a straight corridor (C).

We3-3(3) 15:50-16:05 We3-3(4) 16:05-16:20

Design and Implementation of Interconnecting Design of Transmote: a Modular Self-Reconfigurable IPv6 Wireless Sensor Networks with the Internet Robot with Versatile Transformation Capabilities Guoping Yu and Huasong Min Engineering Research Center of Metallurgical Automation & Measurement Guifang Qiao, Guangming Song, Jun Zhang, Hongtao Sun, Technology of Ministry of Education, Wuhan University of Science & Technology, China Weiguo Wang and Aiguo Song Hongxing Wei School of Instrument Science and Engineering, Southeast University, China Robotics Institute, Beihang University, China • The design and implementation of Transmote • Interconnecting IPv6 WSN with the Internet by access gateway. is presented. • High performance sensor nodes. • Two pairs of IR sensors are used to guide the • Direct access to and control each node by IP remotely. modules to dock. • Multiple Transmote modules can be assembled into I-shaped, T-shaped, O-shaped and X- shaped configurations. The prototype of a single module. • Communication between modules is based on The size is 65mm×72mm× 263mm. ZigBee compliant protocols.

We3-3(5) 16:20-16:35 We3-3(6) 16:35-16:50

Design of a Self-reconfigurable Wireless Network Teleoperation of Humanoid Motion Using System for Modular Self-reconfigurable Robots 3G Communication Network Qian Xu, Qiang Huang, Xuechao Chen, and Zhangguo Yu Guifang Qiao, Guangming Song, Jun Zhang, Hongtao Sun, Jian 1)Intelligent Robotics Institute, Beijing Institute of Technology, China Ge and Weiguo Wang 2)Key Laboratory of Biomimetic Robots and Systems, China School of Instrument Science and Engineering, Southeast University, China 3)Key Laboratory of Intelligent Control and Decision of Complex System, China

• A teleoperation system of Humanoid • Each Transmote module is controlled through Motion .using 3G Communication Network the wireless sensor network. • An on-line motion planning based on • The scale and topology of network will change normalized function. after the deformation of Transmote. • A quick response stability control method • The impacts of multi-hops, LQI, data size and using sensory system. sending interval on communication efficiency Four MSRs transform to an X- are discussed. shaped configuration, and the four • Experiments validate the on-line motion isolated sensor networks (I, II, III, planning and stability control methods. • The function of self-repairing includes both IV) merge into one (V). mechanical repairing and network repairing.

65 IEEE ROBIO 2012 Conference Digest

We3-4: Human Robot Interaction II Session Chairs: Wenhuan Cui and Xia Zhang

Room: Jade 8, 15:20-16:50, Wednesday, December 12, 2012

We3-4(1) 15:20-15:35 We3-4(2) 15:35-15:50

Research on Healthy Subject Gait Cycle Phase Robust Hand Tracking with Refined CAMShift Based at Different Walking Speeds on Combination of Depth and Image Features

Huifeng Zhang, Jinwu Qian, Linyong Shen, Ya'nan Zhang Wenhuan Cui, Wenmin Wang and Hong Liu Laboratory of Intelligent Machine and System, Shanghai University, China, Shenzhen Graduate School, Peking University, China

• This paper verified the changes of gait • This paper studies the problem of tracking global characteristics at different walking speeds. hand motion for HCI/HRI. A figure here • The gait cycle phase was divided based on • The adaptive fusion of motion cue , depth cue movement trends and abnormal gait . and color cue increases robustness of CAMShift. • Designed an experimental method to obtain • The aspect ratio-based blob refinement scheme the gait parameter data. is proposed for segmenting hand from arm. • Used MATLAB-C and statistical methods to • Experiment shows the effectiveness and Robust hand tracking result quantify the changes of gait characteristics. Gait characteristics testing robustness of our approach under complex system situations.

We3-4(3) 15:50-16:05 We3-4(4) 16:05-16:20

The Application of Neural Oscillators as Trajectory Generators to a Wearable Robotic Suit Xia Zhang Chongqing Jiaotong University, China Minoru Hashimoto Shinshu University, Japan • Neural oscillators are used as trajectory generators to a wearable robotic suit. • Symmetrical and unsymmetrical inhibitory connections among neural oscillators are proposed for realizing complex phase differences among multiple joints. • Our proposal is valid for realization of the anti- phase relationship between the left and right hip joints and about 1/4 phase differences between the hip and knee joints on the ipsilateral side. Wearable Robotic Suit

We3-4(5) 16:20-16:35 We3-4(6) 16:35-16:50

Human Position Tracking for Side By Side Human Gesture Recognition through a Kinect Walking Mobile Robot using Foot Positions Sensor Pakorn Udsatid Nattee Niparnan Ye Gu, Ha Do and Weihua Sheng And Attawith Sudsang School of ECE, Oklahoma State University, USA Department of Computer Engineering, Chulalongkorn University, Yongsheng Ou Thailand Shenzhen Institute of Advanced Technology, China

• Human position and direction calculated • A non-intrusive human gesture recognition from detected feet. A figure here system is implemented with a Kinect sensor. A figure here • Use depth data from kinect to find a floor • Hidden Markov Models (HMMs) are used for plane and feet. modeling the dynamics of the gestures. • Extract feet objects which are above the floor • Easy training, person independent and speed plane. flexibility are the features of the detection • Employ a linear Kalman filter to track. system. • Evaluated on a Pioneer robot platform. Gesture based mobile robot control

66 IEEE ROBIO 2012 Conference Digest

We3-5: Rehabilitation Robots I Session Chairs: Weihai Chen and Zhibin Song

Room: Jade 5, 15:20-16:50, Wednesday, December 12, 2012

We3-5(1) 15:20-15:35 We3-5(2) 15:35-15:50 Posture Control Using Prototyping a Parallel Link Arm Driven by Small New Ankle-Foot Assist Device with Pneumatic Actuator for Shoulder Prostheses Stewart Platform Type Parallel Link Mechanisms Onodera Takayuki, Hiroshi Takemura and Hiroshi Mizoguchi Masashi Sekine1, Kento Sugimori2, Tokyo University of Science, Japan Tapio. V. J. Tarvainen2 and Wenwei Yu2 Ming Ding 1Research Center for Frontier Medical Engineering, RIKEN, Japan Chiba University, Japan 2Graduate School of Engineering, Chiba University, Japan • The developed device can measure and assist • The shoulder with a pneumatic the movements of a human ankle joint in six actuator driven parallel link arm was DOF. suggested. • The developed device can be used for • The configuration of the arm was optimized rehabilitation. by the spatial accessibility and manipulability • The developed device uses a Stewart platform indexes. mechanism. • We built the prototype based on the optimal Prototype of a parallel link arm configuration, validated the arm behavior for shoulder prostheses experimentally. Ankle-foot Assist Device

We3-5(3) 15:50-16:05 We3-5(4) 16:05-16:20

Finite element analysis and optimized design of Recognition of Motion of Human Upper Limb using exoskeleton for lower extremity rehabilitation training SEMG in Real Time: Towards Bilateral Rehabilitation 1Zhibin Song, 1Shuxiang Guo, 2Muye Pang and 2Songyuan Zhang Baijun Ding, Jinwu Qian, Linyong Shen and Yanan Zhang, 1Department of Intelligent Mechanical Systems Eng’g, Kagawa University, Laboratory of Intelligent Machine and System, Shanghai University, China 2Japan Graduate School of Engineering, Kagawa University, Japan

• This paper shows the preliminary research on • This paper carries out finite element analysis sEMG-based bilateral rehabilitation for upper limb. and lightweight design of exoskeleton. • In the processing of feature exaction of sEMG, • PRO/E and ANSYS Workbench is used to build Wavelet packet transform (WPT) and the finite element model. autoregressive (AR) model were used to exact the features which are classified through Back- • Static intensity and rigidity and modal analysis. propagation Neural Networks. • Put forward and analyze the optimization • The experimental results show both methods can scheme, and achieve lightweight design. obtain reliable accuracy of motion pattern Three kinds of motion of the The single leg of exoskeleton recognition. Moreover, on the experimental upper limb condition, the recognized accuracy of WPT is higher than that of AR model.

We3-5(5) 16:20-16:35 We3-5(6) 16:35-16:50 Wearable dummy to simulate joint impairment: The Mechanism of Synchronization-Based Control Model for the discontinuous friction resistance Using Neural Oscillators for Walking Assist due to arthritis Shun Ishikawa, Shogo Okamoto, Yasuhiro Akiyama, Xia Zhang Chongqing Jiaotong University, China Kaoru Isogai, Susumu Hara and Yoji Yamada Minoru Hashimoto Mechanical Science and Engineering, Nagoya Univ., Japan Shinshu University, Japan

• A wearable robotic dummy joint Learner of • Synchronization-based control using neural • to simulate disordered joint Cuff physical oscillators is proposed for walking assist. theraphy resistances of human • Mutual joint torque is used as input signal • for supporting physical therapists Flexion to a neural oscillator, and the output of Control method to master manual examination and extension neural oscillator as desired angle. • Simulations and experiments conducted to • A model to simulate discontinuous Motor and encoder explain the mechanism of assist effect and joint friction resistances due to (produce joint Healthy person synchronization action been obtained. diseases such as osteoarthritis was resistances) The validity of using neural oscillator for developed. • Wearable dummy joint walking assist is verified. Experimental device

67 IEEE ROBIO 2012 Conference Digest

We3-6: Machine Learning III Session Chairs: Andrea Soltoggio and I-Ming Chen

Room: Jade 9, 15:20-16:50, Wednesday, December 12, 2012

We3-6(1) 15:20-15:35 We3-6(2) 15:35-15:50

Using movement primitives in interpreting & decomposing complex trajectories in learning-by-doing

Andrea Soltoggio, Andre Lemme and Jochen J. Steil CoR-Lab, Bielefeld University, Germany

• A decomposition algorithm interprets a noisy demonstration as a set of robot-based primitives. • The decomposition occurs incrementally, with the initial identification of the main features of a demonstration. • The method displays generalization capabilities, it is immune to noise and allows the reuse of previously acquired primitive skills.

We3-6(3) 15:50-16:05 We3-6(4) 16:05-16:20

Teaching Coordinated Strategies to Soccer Gas-Source Localization based on Robots via Imitation Reinforcement Learning Saleha Raza, Sajjad Haider and Mary-Anne Williams Jian-Long Wei, Qing-Hao Meng, Member, IEEE, Yang Wang, Ming Artificial Intelligence Lab, Faculty of Computer Science Institute of Business Administration Zeng, Member, IEEE, and Wei Li, Member, IEEE Karachi, Pakistan School of Electrical Engineering and Automation, Tianjin University, China • Multi-robot based (GSL) in turbulence • The paper applies imitation based learning to dominated airflow environments is addressed. learn collaborative strategies for RoboCup • A multi-agent (RL) algorithm is proposed for Soccer Simulation 3D teams. training multiple robots to finish the GSL task. • During imitation, multiple human experts • Simulation results show that multiple robots demonstrate a task which is then learned and could successfully locate the gas source in replicated by autonomous agents. turbulence dominated airflow • Different performance evaluation metrics are Fig. 1One snapshot of the RL • Moreover, the results also demonstrate that the derived to compare the performance of based multi-robot odor searching strategy-sharing RL outperforms the RL which process in Farrell’s plume imitating agents against human-driven agent does not share strategies. environment and the results are very satisfactory.

We3-6(5) 16:20-16:35 We3-6(6) 16:35-16:50 Effect of Low Level Imitation Strategy on an Context-GMM: Incremental Learning of Sparse Autonomous Multi-Robot System Using On-Line Priors for Gaussian Mixture Regression Learning for Cognitive Map Building Arturo Ribes, Jesús Cerquides and Ramón López de Mántaras Abdelhak Chatty, Ilhem Kallel, and Adel M. Alimi IIIA-CSIC, Spain REGIM: REsearch Groups on Intelligent Machine, ENIS, Tunisia Yiannis Demiris Syed Khursheed Hasnain,Philippe Gaussier, and Philipe Laroque EEE, Imperial College of London, UK ETIS: Neuro-cybernetic team, Image and signal Processing, ENSEA, France

• GMM often contain a large amount of • We highlighted the importance of the imitation components for accurate regression. strategy in a Cognitive Multi-Robot System. • Only a small portion of the model is used at • Combining the learning capability with imitation tactic any given time. leads to a positive feedback in the individual and • Context-GMM learns a set of sparse priors population level. used to evaluate only ~10% of the model with • We proved that imitation strategy optimizes the time same accuracy. to explore the resources. Schematic view of different levels • We showed that imitation strategy allows to create a of abstraction cognitive map for the imitator robot (IR) approximately the same shape as the Leader robot (LR).

68 IEEE ROBIO 2012 Conference Digest

We3-7: Man Machine Interface I Session Chairs: Horacio Martinez and Yujin Wakita

Room: Jade 7, 15:20-16:50, Wednesday, December 12, 2012

We3-7(1) 15:20-15:35 We3-7(2) 15:35-15:50

Real-time 3D Visualization in an Open Head movements based control of an intelligent Architecture of a Robotic Application Horacio Martinez, Constantin von Deimling, Rüdieger von wheelchair in an indoor environment Eisenhart-Rothe and Rainer Burgkart Ericka Janet Rechy-Ramirez, Huosheng Hu, Klaus McDonald-Maier Department of Biomechanics, Clinic of Orthopaedics and Sport Orthopaedics of School of Computer Science & Electronic Engineering, University of Essex, U.K. the Technische Universität München, Germany  This paper presents a user-friendly Heinz Ulbrich human machine interface for hands- Institue of Applied Mechanics, Technische Universität München, Germany • Development of an open architecture and a 3D free control of a wheelchair. visualization for a serial robot.  Its two operation modes are based • It helps with the planning process and on head movements. simulation.  Mode 1 uses one head movement to • The controller frequency can be increased up generate command while Mode 2 to 1 KHz. employs four. • A direct force control approach can be  Emotiv EPOC is employed to obtain designed and tested in order to load 3D visualization of the robot the head movement information of specimens. attached to a specimen users.

We3-7(3) 15:50-16:05 We3-7(4) 16:05-16:20 Generic bioelectrical potential signal human- Effect of Dynamic Change of Arm Position on computer interface with electrostimulation Myoelectric Pattern Recognition feedback Noel Segura Meraz and Yasuhisa Hasegawa Jianwei Liu, Dingguo Zhang, Jiayuan He and Xiangyang Zhu Graduate School of Systems and Information Engineering, University of Tsukuba, Japan Robotics Institute, Shanghai Jiao Tong University, China Junji Takahashi Graduate School of engineering, Nagoya University, Japan • Robustness of pattern recognition is important for real life use of myoelectric prosthesis. • Use of residual muscles to control a cursor type • Dynamic Change of Arm Position has adverse interface effect on myoelectric pattern recognition. • Comparison of two different methods to use • A new training strategy is proposed to reduce BEP signal as a control signal this effect. • Use of electrostimulation as a feedback to the • Training the classifier in dynamic condition of user arm increases the robustness of myoelectric Set up of experiment • Comparison of two positions to give feedback pattern recognition.

We3-7(5) 16:20-16:35 We3-7(6) 16:35-16:50

User Evaluation to Apply the Robotic Arm RAPUDA A Real-time Intelligent Shoe-keyboard for for an Upper-limb Disabilities Patient's Daily Life Computer Input Yanbo Tao, Tin Lun Lam, Huihuan Qian and Yangsheng Xu Yujin Wakita, Woo-Keun Yoon, and Natsuki Yamanobe Department of Mechanical and Automation Engineering, Intelligent Systems Research Institute, National Institute of Advanced Industrial The Chinese University of Hong Kong, Hong Kong Science and Technology (AIST), Japan

• To help the patients who have disabilities for • Developed a shoe platform to assist disable their upper limbs because of muscle dystrophy people to tyoe the keyboard. or cervical spine injuries, we have developed A figure here robot arm RAPUDA which can support their • Analyzed the workspace and the best daily life action. movement when human moved. • We make preliminary experiments for these • Used PCA to reduce the force sensor and 3- tasks and consider as training tasks to axis accelerometer. introduce the robot arm to the daily life and • Compared different type of Back propagation some muscle dystrophy patients try the networks. For figure caption (optional), Artificial face rubbing by use arial narrow 20pt font RAPUDA in the hospital. RAPUDA hand towel

69 Thursday, December 13, 2012

Th1-1 Intelligent Systems II Th1-2 Motion Planning I Th1-3 Tele-robotics and Networked Robots Th1-4 Multi-Robot Systems I Th1-5 Rehabilitation Robots II Th1-6 Robot Control I Th1-7 Man Machine Interface II Th2-1 Intelligent Systems III Th2-2 Motion Planning II Th2-3 Vision/Image Processing III Th2-4 Multi-Robot Systems II Th2-5 Rehabilitation Robots III Th2-6 Robot Control II Th2-7 Sensors I Th3-1 Modular Robot Systems Th3-2 Motion Planning III Th3-3 Vision/Image Processing IV Th3-4 Multi-Robot Systems III Th3-5 Search Rescue Robots Th3-6 Robot Control III Th3-7 Sensors II

IEEE ROBIO 2012 Conference Digest

Th1-1: Intelligent Systems II Session Chairs: Wenrong Wu and Grienggrai Rajchakit

Room: Jade 1, 10:20-12:05, Thursday, December 13, 2012

Th1-1(1) 10:20-10:35 Th1-1(2) 10:35-10:50

Decision Model for Determing the Behavior of Pedestrians Approaching from Both Sides of a Blind Corner by Considering Tactics and Information Uncertainty Tamao Okamoto*&** and Yoji Yamada* Department of Science and Engineering, Nagoya University, JAPAN * Exponential stability of switched Panasonic Corporation, JAPAN ** linear systems with

• Game Theory based Decision Model Decision Model Interval time-varying delays reproduce human behaviors by tactics • Decision Model was expanded using Grienggrai Rajchakit Theory of Evidence for considering Major of Mathematics, Faculty of Science information uncertainty Maejo University, Chiangmai, 50290 Thailand. Which • Proposed Model was verified through Course is Chosen ? simulation In/Out

• Contribution to Traffic Management Collision× for Mobile Robots & Pedestrians In /Out ×Detour

Th1-1(3) 10:50-11:05 Th1-1(4) 11:05-11:20

Human Recognition with a Hardware-Accelerated Primitive action learning using fuzzy neural networks Multi-Prototype Learning and Classification System Y. Gatsoulis, I. Siradjuddin and T.M. McGinnity Indra Bagus Wicaksono, Fengwei An, and Hans Juergen Mattausch University of Ulster, UK Research Institute for Nanodevice and Bio Systems Hiroshima University, Japan  Affordances learning has always been a top research topic and is still open

 Using a Takaki-Sugeno Fuzzy Neural • A flexible hardware-accelerated learning and Network (TSFNN) for learning how to classification system which surmounts previous push rolling objects systems’ complexity issues.  Benchmarked with a linear regression • System applied to human recognition to model demonstrate capabilities.  Experiments with a physical PR2 robot • Accelerated learning and classification speeds through FPGA implementation  Results showed the ability of the TSFNN to accurately and robustly push rolling objects

Diagram For Learning Architecture  Better performance than the linear model

Th1-1(5) 11:20-11:35 Th1-1(6) 11:35-11:50 HYDRAULIC PROSTHETIC KNEE MECHANISM WITH NEURAL NETWORK INTERPHASE A Precision Robotic system for K. Chaitanya1 Y.Kalyan Chakravarthy2 A.Srinath2* K.Rajasekhara Rao3 Abstract--Prosthetic knee mechanism helps in 3-D Micro-Assembly framing a better prosthetic knee for those individuals with knee amputations in the portions Wu wenrong andYu dai-hai and Du kai above knee joint. Laser Fusion Research Center, CAEP Sichuan, china, The primary objective of this mechanism is that it Qiu zurong should withstand the human loads, be flexible to the movements and also be able to mingle with the State Key Lab of Precision Measuring Technology & Instruments, Tianjin amputated part in the similar manner as that of the University, china natural knee. The said Mechanism is developed based on • 3-D micro-assembly hydraulics principle which avoids wear and tear of the material and the friction within it. Conclusions: • precision motion control A figure here The 3D model has been created using Pro-Engineer The proposed neural networks articulation can be easily adapted by modeling software. suitable selection of Neural Controllers and its design and • micro gripper incorporation into an Embedded System, that can thoroughly govern The main objective of the paper is to develop a the designed hydraulic knee from the flex forces and reflex forces knee mechanism that is more flexible, economical, generated by the limbs and joints at the place where this prosthetic • micro manipulation. efficient, easy to manufacture than the existing hydraulic knee is connected to the human thigh ( leg). models. Future Scope: Implementation of a neural controller ca be described that is used The dimensions considered for modeling the as an arbitrator or selector of a number of behaviors. A genetic mechanism are taken based on the size of the algorithm can be used to evolve a neural network that satisfies the amputated knee. requirements. Precision robotic system The Neural Networks articulation is proposed which Reference: can be adapted using an Integrated Circuit or an Design and analysis of prosthetic knee joint by Prof. W. L. Cleghorn& J. Embedded System for proper functionality Andrysek. achievement of the proposed hydraulic knee joint. GURNEY, K. Neural Nets, UCL Press, London

71 IEEE ROBIO 2012 Conference Digest

Th1-2: Motion Planning I Session Chairs: Chuangqi Wang and Wenfu Xu

Room: Jade 2, 10:20-12:05, Thursday, December 13, 2012

Th1-2(1) 10:20-10:35 Th1-2(2) 10:35-10:50

Path Updating Tree based Fast Path Planner Singularity-free Path Planning of Dual-arm for Unpredictable Changing Environments Space Robot for Keeping the Base Stabilized Chuangqi Wang and Hong Liu* Wenfu Xu, Houde Liu, Yu She, Enginnering Lab on Intelligent Perception for Internet of Things(ELIP), and Bin Liang, Member, IEEE Shenzhen Graduate School, Peking University, China Shenzhen Graduate School, Harbin Institute of Technology Bin Chen Harbin Institute of Technology at Weihai, China • Modeling of Dual-arm Space Robot System • Path Updating Tree (PUT) is proposed to explore the environment globally by • The Base Centroid Virtual Manipulator updating the extended nodes. Model • The fast path planner with two-level is • Coordinated Path Planning for introduced to separate exploring Stabilizing the Base During Target globally with exploiting locally guided by the PUT to distribute computational • Simulation Study resource more properly. • Discussion and Conclusion A two-level path planner with PUT

Th1-2(3) 10:50-11:05 Th1-2(4) 11:05-11:20

A Cell Based Voronoi Roadmap for Motion Planning A Floyd-Genetic Algorithm Based Path Planning of Articulated Robots Using Movement Primitives System for Mobile Robots in Laboratory Automation C. Kohrt, A. G. Pipe, J. Kiely and R. Stamp University of the West of England, Bristol, UK Hui Liu a, Norbert Stoll b, Steffen Junginger a and Kerstin Thurow b G. Schiedermeier a. Institute of Automation, University of Rostock, Germany University of Applied Sciences Landshut b. Center for Life Science Automation (celisca), Germany

• The applied cell based Voronoi • A standard robot path planning system (RPPS) roadmap algorithm is fast and realizes is developed. online robot path planning. • A hybrid path planning method based on map • The combination of Cartesian and theory and artificial intelligence is proposed. joint space efficiently utilizes existing information. • All necessary technologies of laboratory mobile robotics are considered. • The elastic net creates the trajectory A laboratory mobile robot which consists of linear and circular • A real robot transportation case using the movement primitives. presented RPPS is demonstrated.

Th1-2(5) 11:20-11:35 Th1-2(6) 11:35-11:50

Path Planning Algorithm for Space Manipulator Dynamics Analysis for a Five DOF Serial-parallel with a Minimum Energy Demand Manipulator with RP-(2-RRU/1-RUU) Structure Chao Hou5, Peng Zhang1,2,4*, Ying Hu1,2, Jianwei Zhang3 Wing Kwong Chung, and Yangsheng Xu, 1Shenzhen Institutes of Advanced Technology, Chinese Academy of Sciences Shenzhen Institute of Advanced 2The Chinese University of Hong Kong, Hong Kong, China Integration Technology, Chinese Academy of Sciences / 3TAMS, Department of Informatics, University of Hamburg, Hamburg, Germany The Chinese University of Hong Kong 4Shenzhen Key Laboratory of Precision Engineering 5Harbin Institute of Technology, Shenzhen Graduate School • Propose a methodology to plan a minimum • The modal analysis and the harmonic response energy demand path for space manipulators analysis are performed, and the dynamic • Develop a new genetic operator, Planar-NN characteristics of the manipulator are derived. • The inverse dynamics model is built to find the • Enhance the final solution in terms of fast relationship between the actuators’ input forces convergent rate with a lower energy demand and the output motion of end-effector. • The dynamics simulation is carried out by MATLAB, which verifies the correctness of 5-DOF Serial-parallel dynamics modeling. Manipulator

72 IEEE ROBIO 2012 Conference Digest

Th1-3: Tele-robotics and Networked Robots Session Chairs: Qian Xu and Abderrahmane Kheddar

Room: Jade 3, 10:20-12:05, Thursday, December 13, 2012

Th1-3(1) 10:20-10:35 Th1-3(2) 10:35-10:50

Development of a Teleoperation System With a Tele-Impedance Control of a New Workspace Spanning Technique Variable Stiffness Prosthetic Hand Elif Hocaoglu and Volkan Patoglu Mohamed Mamdouh Faculty of Engineering and Natural Sciences Mechatronics, Egypt Japnan University of Science and Technology , Egypt Sabancı University, Istanbul, Turkey Ahmed Ramadan Computer Science and control, Tanta University, Egypt • A variable stiffness prosthetic hand and its • Bilateral teleoperation system consist of surface electromyography interface for tele- phantom premium as a master and Mitsubishi impedance control are presented. robot arm is developed. • System enables the amputee to modulate the • New workspace mapping method is proposed impedance of the prosthetic limb to properly to map the workspace between the two match the requirements of the task. dissimilar robot. • The impedance regulation takes place naturally, while position of the hand prosthesis is regulated • Experimental tests show a good tracking Developed Master/Slave intentionally. characteristic of the system. teleoperation system Test setup for the variable • The proposed interface does not require long stiffness prosthetic hand • 6 DOF F/T sensor is used to sense the training, and provides amputee with ease of use. interaction force with the enviroment

Th1-3(3) 10:50-11:05 Th1-3(4) 11:05-11:20

A Remote Control System with Combination of Compliance Estimation during Bilateral Manual and Automatic Control Teleoperation of a Robotic Arm Using Gripping Force

Kazuma Shimojo, Yasuo Hayashibara and Ken Tomiyama Remo Pillat and Arjun Nagendran The Department of Advanced Robotics, Chiba Institute of Technology, Japan Synthetic Reality Lab, University of Central Florida, USA

• Proposed a new method to combine manual • Estimate compliance properties of object and automatic controls for remote-controlled while probing with a robotic arm robots • Ensure non-destructive behavior and limit • Adopted the gripping force of the joystick for contact forces through admittance control the determination of mixing ratio • Compare noise-resiliency of RANSAC- • Developed a simulator for evaluation of the based method and popular SPRLS method proposed method • Compared four methods to show superiority of Experimental system to evaluate • Promising results for simulated and real- the proposed method the proposed method in many aspects world experiments

Th1-3(5) 11:20-11:35

On-line Operator Skill Assessment for Telerobot Operation Using Electroencephalo-graph (EEG) Shuang Liu and Ning Xi Department of MBE, City University of Hong Kong, Hong Kong Yunyi Jia Department of ECE, Michigan State University, USA • This paper presents an on-line skill assessment method in the teleoperation, based on a new concept named by skill of teleoperator (SoT), which represents the skill level of the human teleoperator from the view angle of the brain states. The basic idea is that the teleoperators with different skill levels have different brain states during the teleoperation. The skill level of the teleoperator can be easily assessed on line by wearing an Emotiv headset during the Skill Assessment for teleoperation. Teleoperation Using EEG

73 IEEE ROBIO 2012 Conference Digest

Th1-4: Multi-Robot Systems I Session Chairs: Dong Sun and Kuniaki Kawabata

Room: Jade 8, 10:20-12:05, Thursday, December 13, 2012

Th1-4(1) 10:20-10:35 Th1-4(2) 10:35-10:50

Wave-Transformation-Based Control Law for Distributed Multirobot Shape Control with a Multilevel- Teleoperation with Large Time-varying Delays based Topology and Market-based Auction Algorithm Xiao Yan, and Dong Sun Fangping Yang, Hongyi Li and Yuechao Wang State Key Laboratory of Robotics, Shenyang Institute of Automation, Chinese Department of Mechanical and , Academy of Sciences, Shenyang,China City University of Hong Kong, Hong Kong

• a solution forTele-operation with Large Time- varying Delays. • Robots converge to a desired region • adding a filter inside the communication and construct a desired shape channel based on the wave transformation. • A multilevel-based topology design for • conduct several experiments with a six-degree multirobot shape control of freedom master/slave system. • Market-based auction algorithm is Wave-Transformation-Based used for task assignment Control law

Th1-4(3) 10:50-11:05 Th1-4(4) 11:05-11:20

Development of a Cricket Interaction System Distributed Motion Control Algorithm utilizing Mobile Robot for Behavioral Data Collection for Fission Behavior of Flocks Kuniaki Kawabata RIKEN, Japan/Inst. of Artificial Intelligence and Robotics, XJTU, China Xiaokang Lei, Mingyong Liu, Wenbai Li and Panpan Yang Hitoshi Aonuma, Koh Hosoda School of Marine Engineering, RIES, Hokkaido Univ., Japan Osaka Univ. Japan Northwestern Polytechnical University, China Jianru Xue Inst. of Artificial Intelligence and Robotics, Xi’an Jiaotong Univ., China • Our goal is to develop a distributed motion • Cricket-micro mobile robot interaction system control algorithm for the anonymous swarm for behavioral experiment. IR transmitter system to generate fission behavior. CCD camera • We propose a modified flocking algorithm by • Manipulatable experimental conditions Microphone integrating a local following interaction with • Real-time visual tracking of cricket and micro attraction, repulsion and alignment behaviors. mobile robot Arena • Simulation results show that the group-level • On-line data collection of video, trajectories, fission behavior emerges as a spontaneous sound during interaction experiment result of flock responding to multiple stimulus.

Th1-4(5) 11:20-11:35 Th1-4(6) 11:35-11:50 Decentralized task allocation for Cooperative Control and Communication heterogeneous agent systems with constraints Optimization for Networked Multi-agent Systems on agent capacity and critical tasks with Nonholonomic Constraints Weirong Liu and Feng Zhou Giulio Binetti, David Naso and Biagio Turchiano School of Information Science&Engineering, Central South University, China Department of Electrical and Electronic Engineering (DEE), Jing Wang Technical University of Bari, Italy Computer Engineering Program, Bethune-Cookman University, USA

tasks • A formation control law for multi-agent system This paper proposes a decentralized allocation with nonholonomic constraints is proposed. A figure here algorithm that builds an initial assignment • A best distance between agents determined by through a market-based approach and resolves communication quality is introduced to get a conflicts with a consensus procedure based on critical tasks trade-off between coverage and reliable local peer-to-peer communications between communication. neighboring robotic agents, considering systems robotic • .a model-free extremum seeking algorithm is with functional heterogeneity, resource agents used to find the best distance among agents. constraints and guarantee of assignment for Example of assignment critical tasks. Formation Result

74 IEEE ROBIO 2012 Conference Digest

Th1-5: Rehabilitation Robots II Session Chairs: Tim Lueth and Nianfeng Wang

Room: Jade 5, 10:20-12:05, Thursday, December 13, 2012

Th1-5(1) 10:20-10:35 Th1-5(2) 10:35-10:50

Kinematic Design of a Parallel Ankle Rehabilitation Analysis for the foot board lower limb Robot for Sprained Ankle Physiotherapy rehabilitation robot Yongfeng Wang and Guoru Zhao Yongqiang Wu, Yanan Zhang, Linyong Shen and Jinwu Qian Key Laboratory for , Shenzhen Institutes of Advanced Laboratory of Intelligent Machine and System, Shanghai University, China Technology, Chinese Academy of Sciences, Shenzhen 518055, P. R. China

• It presented a novel parallel ankle rehabilitation • This paper describes a new foot board lower robot, and achieved the kinematic solution and A figure here limb rehabilitation robot. simulation analysis. • Use NDI OPTOTRAK system to measure the • The designed ankle rehabilitation robot had the trajectory and angle of ankle. simple and compact structure, automatic • Have done the kinematics analysis of robot. adjustment function. • Set a model in • This robot was properly to meet the range of MATLAB/SIMULINK module and simulate. 6-dof foot board lower limb motion for the normal ankle, and helped rehabilitation robot patients to take various kinds of ankle rehabilitation exercising. The parallel ankle rehabilitation robot

Th1-5(3) 10:50-11:05 Th1-5(4) 11:05-11:20

Design of a Compliant Prosthesis Finger by A Robust Adaptive Neural Controller to Drive a Knee Joint Actuated Orthosis Structural Topology Optimization S. Mefoued, M. E. Daachi, B. Daachi,S. Mohammed and Y. Amirat

Nianfeng Wang and Xianmin Zhang University of Paris Est Creteil - (UPEC) Guangdong Province Key Laboratory of Precision Equipment and LISSI Laboratory, 94400, Vitry Sur Seine, France Manufacturing Technology, South China University of Technology, China Design of an actuated orthosis intended to restore lower limb movements of dependent persons. • As to the mechanical design, mechanical structures Robust adaptive control based on a Multi-Layer for robotic hand based on the ``compliant Perception Neural Network (MLPNN) mechanism'' concept proves to be quite satisfactory in terms of structural simplification, ease of A neural identification is used to extract the principal components of the MLPNN input vector. manufacturing and assembly. Actuated orthosis ”EICoSI” worn by a • As to the design method, an efficient geometric The MLPNN is used to compensate the dynamic subject in sitting position. effects arising from the inter-action between the representation scheme using fat Bezier curve to human lower limb and the orthosis. directly generate smooth boundary structures has Evaluation of controllers’ performances in terms of tracking errors and robustness been used. against external perturbations. Sketch of a hand

Th1-5(5) 11:20-11:35 Th1-5(6) 11:35-11:50 Virtual Therapist: A Phantom Robot-Based Haptic System for Personalized Post-Surgery Finger Rehabilitation Zhe Xu*, Rebecca Fiebrink†, and Yoky Matsuoka* *Department of Computer Science & Engineering, , WA, USA † Department of Computer Science, Princeton University, NJ, USA • Design of a haptic system for improving patient motivation and reducing barriers to therapist customization of the rehabilitation system. • To reduce contact-induced infection, an anatomically correct finger model is derived for coupling the fingertip of a patient. • Personalized software interface and fingertip connector can accommodate patients with different treatment plans and physical Hardware and software interfaces conditions. of the Virtual Therapist

75 IEEE ROBIO 2012 Conference Digest

Th1-6: Robot Control I Session Chairs: Junjian Lin and Zhicheng Qiu

Room: Jade 9, 10:20-12:05, Thursday, December 13, 2012

Th1-6(1) 10:20-10:35 Th1-6(2) 10:35-10:50

Fuzzy PD Compliance Control of 6 Degrees-of- End Edge Feedback and RBF Neural Network Based Freedom Robot Using Disturbed Force Sense Vibration Control of Flexible Manipulator Junjian Lin and Tie Zhang Zhicheng Qiu , Biao Ma, Xiangtong Zhang Mechanical & Automotive Engineering, SCUT, China School of Mechanical and Automotive Engineering , South China University of Technology, Guangzhou, China

• A vision feedback based active vibration • Show the system schematic. Use position Computer and control board control system is proposed . Voltage amplifier motors, and disturb the force sense. • Active vibration control is conducted by • Controller design. A fuzzy PD controller with a PZT actuator employing RBF neural networks. lowpass filter is proposed. Light source Flexible beam

• Experiment the proposed controller and the • The simulation and experimental results lens traditional PD controller. demonstrate that the proposed system and CCD camera • An evaluation method is proposed and used to algorithm can suppress the vibration of the compare with the two methods above. flexible manipulator effectively.

Th1-6(3) 10:50-11:05 Th1-6(4) 11:05-11:20

Operator based robust control for robot Design and implementation of a vehicle dynamics control manipulator using robust right system by means of torque vectoring for an autonomous coprime factorization vehicle Shuhui Bi and Guangliang Liu Contents : Shandong Provincial Key Laboratory of .Concept for the integration of a Vehicle- Robot and Manufacturing Automation Technology Dynamics Control System in AGVs Institute of Automation,Shandong Academy of Sciences, China .Implementation in a scaled vehicle as a Mingcong Deng mechatronic platform for an AGV Department of Electrical and Electronic Engineering Tokyo University of Agriculture and Technology, Japan .Evaluation in two Use-Cases

[10*°/s] • Right coprime factorization of one class of robot dynamics like nonlinear systems is considered. • An operator based control design structure for [20ms] robot manipulator is proposed. Operator based robust control demandedSoll-Gierrate yaw rate Ist-Gierratereal yaw rate Ist-Gierratereal yaw rate(TV) (TV) • Simulation is done to validate the effectiveness 08.12.2010 of the proposed control design. Sebastian Reuter

Th1-6(5) 11:20-11:35 Th1-6(6) 11:35-11:50

A Fault Tolerant Adaptive Control for Robot Combining Stereo Vision and Fuzzy Image Based for Autonomous Object Manipulators Grasping Using a 6-DoF Manipulator Giuseppe Muscio and Francesco Pierri Le Duc Hanh and Chyi-Yeu Lin School of Engineering, University of Basilicata, Italy Mechanical Engineering, National Taiwan University of Science and Technology, Taiwan • Robotic manipulator with faults affecting the joint actuators. • A combination between Stereo Vision and • Active fault tolerant approach in two phases: Fuzzy Image Based Visual Servoing is applied. • observer based fault detection and isolation, • The advantage helps to save the precious time • controller reconfiguration. and computational cost when applying in various environments. • Inverse dynamic robust controller reconfigured via an adaptive term compensating the effect • The grasping technique can be duplicated in of the fault. different environments easily • Fault estimation by using universal interpolator. • The same methodology can be extended to grasp objects with other geometric shapes

76 IEEE ROBIO 2012 Conference Digest

Th1-7: Man Machine Interface II Session Chairs: Ortiz Jesus and Jeffrey Too Chuan Tan

Room: Jade 7, 10:20-12:05, Thursday, December 13, 2012

Th1-7(1) 10:20-10:35 Th1-7(2) 10:35-10:50

Design of Unit for Haptic Device on Foot Hand Detection Using Multi-Resolution HOG Features Yasunobu Masuda, Takehito Kikuchi Member, IEEE, Wataru Kobayashi, Tetsu Mitsumata and Suguru Ohori Yanguo Zhao, Zhan Song and Xinyu Wu The Graduate School of Science and Engineering, Shenzhen Institutes of Advanced Technology, CAS/CUHK Yamagata University, Japan • Proposed a multi-resolution HoG based representation from the proposed GOI, which • We developed an electromagnet as a testbed helps improving robustness of the features. of magnetic field sensitive elastomer and • Proposed to use the piece-wise linear evaluated it. decision function based cascade classifier for object classification, which helps boosting the efficiency . • Experimental results with practical scenarios • We designed a unit for haptic device on foot show its robustness as well as the with MSE. improvement in detection accuracy and efficiency.

Th1-7(3) 10:50-11:05 Th1-7(4) 11:05-11:20

Smart Devices in Robot-Assisted Laser Microsurgery: Multimodal Human-Robot Interaction with Chatterbot System: Towards Ubiquitous Tele-Cooperation Extending AIML Towards Supporting Embodied Interactions Jesús Ortiz, Leonardo S. Mattos and Darwin G. Caldwell Jeffrey Too Chuan Tan and Tetsunari Inamura Department of Advanced Robotics, Istituto Italiano di Tecnologia, via Morego, 30, 16163 National Institute of Informatics, Japan • Laser microsurgery requires a high Feng Duan level of precision Nankai University, P. R. China • Classical micromanipulator requires extensive training and the Objective: To realize multimodal HRI in embodied performance is highly dependent virtual environment based on a simple chatterbot This is apple. from the surgeon skills • Using the virtual scalpel the dialogue system surgeon can control the laser with a. Gesture Inputs – User’s pointing target is a pen drawing on the live video • Using a smart device we can as the predicate in the conversation increase the possibilities of the b. Emotional Expressions – The tag can system adding tele-cooperation be used to set different moods capabilities What is this? • The precision of the smart device Virtual scalpel with mobile tele-cooperation capabilities: The c. Robot Interactive Learning – is is similar to the micromanipulator graphics pen controls in real-time the aiming and activation of the surgical laser by touching the live microscope video extended to interactively learn new knowledge Multimodal HRI with but with less maximum error shown on either interactive display. d. Interactive Learning towards Symbol Grounding Embodied Interactions

Th1-7(5) 11:20-11:35 Th1-7(6) 11:35-11:50 Intuitive Operation of A Wheelchair Mounted Design and Implementation of Behaviour Detection Robotic Arm for The Upper Limb Disabled: System for the Elderly based on Smart Phone The Mouth-only Approach Min Dong, Zhi Hou, Zhitian Liu,Sheng Bi, Peidong Lin, Wei Wang, Zeming Zhang, Yuki Suga, Hiroyasu Iwata and Shigeki Liangchang Zhu, Jiayi Xin and Liangpeng Liu Sugano School of Compuer & Engineering, South China University of Technology Department of Modern Mechanical Engineering, Waseda University, Japan • A behaviour detection system with • A new interfacing device that enables intuitive a sensor board (iNEMO V2) and operation of an assistive robotic arm. A figure here the iNEMO SDK (software A figure here • Three kinds of physical signals: jaw movement, development kit) is proposed in this breathing strength and head movement are paper , which utilizes the captured and utilized using this device. accelerometer, micro-sensors, • The device is designed to operate in a mouth- digital data processing as well as only manner, thus enables usage for the upper wireless technologies. The system limb disabled For figure caption (optional), practically simulates a real environment for the intended case. • A compact and light-weight design use arial narrow 20pt font

77 IEEE ROBIO 2012 Conference Digest

Th1-7: Man Machine Interface II (cont.) Session Chairs: Ortiz Jesus and Jeffrey Too Chuan Tan

Room: Jade 7, 10:20-12:05, Thursday, December 13, 2012

Th1-7(7) 11:50-12:05

Real-time Hand-Writing Tracking and Recognition by Integrated Micro Motion and Vision Sensors Platform Shengli Zhou and Ronald Chung Department of Mechanical and Automation Engineering, The Chinese University of Hong Kong, Hong Kong SAR Guanglie Zhang and Wen J. Li Department of Mechanical and Biomedical Engineering, City University of Hong Kong, Hong Kong SAR Jay Y. J. Liou Hong Kong Applied Science and Technology Research Institute, Hong Kong SAR • An algorithm has been developed for simultaneously hand-writing tracking and recognition • The100 Hz inertial measurements and 5 Hz visual measurements are fused with EKF • Reconstructed trajectories are recognized by Tracking and recognition using DTW with an accuracy of 88.46% results.

78 IEEE ROBIO 2012 Conference Digest

Th2-1: Intelligent Systems III Session Chairs: Guanglie Zhang and Tetsuya Wada

Room: Jade 1, 13:30-15:00, Thursday, December 13, 2012

Th2-1(1) 13:30-13:45 Th2-1(2) 13:45-14:00

Development of Virtual Reality Bike with A New Prismatic Series Elastic Actuator with Compact Size and Cylindrical MR Fluid Brake High Performance Nicholas Paine and Luis Sentis Takehito Kikuchi, Keigo Kobayashi, and Masao Sugiyama Human Centered Robotics Lab, University of Texas Graduate School of Science and Engineering, Yamagata University, Japan Austin, TX, USA

• Novel mechanical design for We have developed the virtual reality compactness bike (VR Bike) with a cylindrical MR High motor voltage for increased fluid brake. • power output The aim of this paper is to describe • Model-based force and position (1) the development and integration controllers for accurate and stable of the total system of the VR Bike, regulation and • Strong experimental results (2) the implementation of the VR Bike System showing 94W/kg and 77% controller for the pedaling torque in mechanical efficiency the straightway driving with slopes.

Th2-1(3) 14:00-14:15 Th2-1(4) 14:15-14:30

A Real-time Velocity Estimation using Motion Design of a Network-based Mobile Gait Blur in Air Hockey Rehabilitation System Shayan Rezvankhah, Amirali. Bagherzadeh Wenlong Zhang, Xiuming Zhu, Song Han, Electrical and Computer Eng., University of Tehran, Iran Nancy Byl, Aloysius K. Mok, and Masayoshi Tomizuka Hadi Moradi, Babak Araabi UC Berkeley, UT Austin, and UC San Francisco Electrical and Computer Eng., University of Tehran, Iran United States of America

• a real-time approach for the air hockey puck’s • A Network-based Rehabilitation System for velocity estimation based on motion blur is Enhanced Mobility and Tele-rehabilitation A figure here presented. • A New Wireless Protocol (MBStarPlus) for • low resolution image is used to detect the puck Real-time Communication with a High • Fourier transform is used on the full resolution Sampling Rate and Low Power Consumption image to extract the direction of the motion. • A Modified LQG Controller with a Disturbance • The roundness of the ball and the length of the Observer (DOB) to Control the Actuator blur are finally used in order to estimate the • Experimental Results and Clinical Test Design speed.

Th2-1(5) 14:30-14:45 Th2-1(6) 14:45-15:00 Robot Aided Object Segmentation Based on Micro Vision Based Cell Motility Analyzing Kinect without Prior Knowledge Algorithm by Optically-Induced Dielectrophoresis Guanglie Zhang1, Long-Ho Chau2, Mengxing Ouyang1, and Wen J. Li1* Wenzheng Chi School of Control Science and Engineering, Shandong University, China 1City University of Hong Kong, and 2The Chinese University of Hong Kong Max Q.-H. Meng and Xijun Chen  Our group has performed the self- Department of Electronic Engineering, The Chinese University of Hong Kong, induced rotation of Melan-A cells in China the optically-induced dielectrophoretic • A new object segmentation method is proposed system. The cells could rotate during based on 3-D information and robot stationary and translate in a non- manipulation. uniform E-Field. • Kinect to obtain 3-D information simplifies the  The optical flow method in the segmentation process and improves its microscopic video sequences is used performance. to estimate the 3D motion in the dielectrophoresis force fields, • Robot arm WAM with a larger payload enables including rotation and translational our platform to deal with heavier objects. motion. • Experiments in three cases have proved the effectiveness of our method. Object segmentation process

79 IEEE ROBIO 2012 Conference Digest

Th2-2: Motion Planning II Session Chairs: I-Ming Chen and Chris Smith

Room: Jade 2, 13:30-15:00, Thursday, December 13, 2012

Th2-2(1) 13:30-13:45 Th2-2(2) 13:45-14:00

Dynamics based Time-Optimal Smooth Motion Global Path Planning for Mobile Robot Based on Planning for the Delta Robot A* Algorithm and Genetic Algorithm Liang Zhang, Huasong Min and Haojun Huang Yunqiang Zhang1, Ruining Huang1, Yunjiang Lou1, 2 Engineering Research Center of Metallurgical Automation and Measurement and Zexiang Li Technology of Ministry of Education, Wuhan University of Science and 1 Harbin Institute of Technology Shenzhen Graduate School, China Technology, China 2 Hong Kong University of Science and Technology, China Hongxing Wei Robotics Institute, Beihang University, China • Motion is smooth and suitable for high speed applications to avoid vibration. A figure here • Global Path Planning in Grid Map. • Time-optimal problem is transformed into a • A* Algorithm and Genetic Algorithm. three-variable optimization problem. • Shorter Path and Not Limited to The Center of • Rapidly obtain a solution that guarantees the The Grid . smoothness of torques. • Higher Efficiency and Faster Search Speed. • The complete model of delta robot is used. Mobile robot’s working environment

Th2-2(3) 14:00-14:15 Th2-2(4) 14:15-14:30

Wheeled-tip Object Manipulation: Modeling and Offline Motion Planning of Two-Robot Motion Planning of Throwing an Object Coordination in Exhaust Manifold Welding

Sina Parastegari, Majid Nili Ahmadabadi, Ehsan Noohi and Hadi Fan Ouyang, Tie Zhang, Can Ying School of Mechanical & Automotive Engineering, Moradi South China University of Technology School of Electrical and Computer Engineering, University of Tehran, Iran

• Modeling the task of throwing the objects as a Two advantages of two 6 DoF robot coordinated non-prehensile manipulation task. welding can be concluded : A figure here • Presenting a PRM based motion planning • is especially good for irregular part due to the algorithm. holding position of clamping robot can be • Taking advantages of the Wheeled-tip arbitrary . structure in the throwing task. • can be suitable for continuous welding of multi- spatial seams without changing the holding The robot transfers the object from position . Two-robot coordinated welding the initial position to the throwing position simulation (Seam 1, 2, 3) and throws it toward the goal position

Th2-2(5) 14:30-14:45 Th2-2(6) 14:45-15:00

Periodic Trajectory Generation and Tracking Control Measurement Framework of Partial Cage Quality for a Class of Underactuated Mechanical Systems Hongtai Cheng1 and Heping Chen1, Xiaohua Zhang2 Teesit Makapunyo, Thanathorn Phoka, Peam Pipattanasomporn, Bingtuan Gao3 and Hongjun Chen4 Nattee Niparnan and Attawith Sudsang 1.Texas State University-San Marcos, USA; 2.Dalian University of Technology, China; Department of Computer Engineering, Chulalongkorn University, Thailand 3.Southeast University, China; 4. Harbin Institute of Technology, China • This work aims to realize the arbitrary state • The concept of partial cage space point steering control for the UMS. • Basic idea of quality of partial cage • A target-oriented periodic trajectory generation • The framework to measure partial cage quality technique is proposed to generate trajectories and its preliminary experiment that pass through the desired position. • Criteria are presented to determine the periodic property of the generated trajectories. • The proposed Lyapunov based controller can be implemented on a class of UMS to realize complex tasks and extend their applications

80 IEEE ROBIO 2012 Conference Digest

Th2-3: Vision/Image Processing III Session Chairs: Marcel Haselich and Fei Chao

Room: Jade 3, 13:30-15:00, Thursday, December 13, 2012

Th2-3(1) 13:30-13:45 Th2-3(2) 13:45-14:00 Parallelized Energy Minimization for Real-Time Robust Visual Tracking Based on Gabor Feature Markov Random Field Terrain Classification in and Sparse Natural Environments Weiguang Li, Yueen Hou, Huidong Lou and Guoqiang Ye Marcel Häselich, Simon Eggert and Dietrich Paulus School of Mechanical and Automotive Engineering, South China University of Active Vision Group, University of Koblenz-Landau Technology, China Universitätsstr. 1, 56070 Koblenz, Germany robots.uni-koblenz.de • Proposes a novel approach to deal with the problem of visual tracking in • Terrain classification for 3D laser range finder challenging situations data in real-time using Markov random fields • Use the Gabor wavelet to represent the features • Recall ratios of about 90% for detecting • Use the sparse representation to linearly combine the candidate target obstacles and drivable areas • Use the incremental learning method to update the template dictionary • Multicore parallelization and general-purpose computing on graphics processing unit For figure caption (optional), • Acceleration of energy minimization by a use arial narrow 20pt font factor of 10

Th2-3(3) 14:00-14:15 Th2-3(4) 14:15-14:30

Water Detection with Segmentation Guided Forward Obstacle Detection System Dynamic Texture Recognition by Stereo Vision

1,2 1 1 Hiroaki Iwata Pedro Santana and Ricardo Mendonça and José Barata Department of Physics, Tokyo Institute of Technology, Japan 1 CTS-UNINOVA, Portugal 2 University Institute of Lisbon (ISCTE-IUL), Portugal Keiji Saneyoshi Radiation Research Center, Tokyo Institute of Technology, Japan

• Chaotic nature of rippled water’s dynamic • A forward obstacle detection system using only texture exploited for water/land segmentation; stereo vision has developed. • Chaotic nature of rippled water assessed based • This system runs in real time by employing a on a set of entropy-based measures; FPGA-based stereo processing system. • Locally flat water regions handled with a • Various experiments were done to evaluate the segmentation-guided label propagation method; performance of the proposed system. • A ROC-based analysis over 12 DynTex videos confirms a TPR/FPR trade-off of 0.95/0.1.

Th2-3(5) 14:30-14:45 Th2-3(6) 14:45-15:00

A Developmental Constraint Driven Approach to Online Motion Segmentation using Robotic Hand-eye Coordination Spatially-constrained J-linkage in Dynamic Scene

Jungwon Kang, Sang Un Park and Myung Jin Chung Fei Chao, Haixiong Lin, Min Jiang and Changle Zhou Department of Electrical Engineering, KAIST, Republic of Korea Department, Xiamen University, P. R. China

• Sensory-motor learning plays an important role • Clustered trajectories in dynamic scenes. in robotic cognition. A figure here • Two constructive neural networks are applied • Presented spatially-constrained to simulate human brain loops to handle J-linkage for clustering trajectories robotic hand-eye coordination. by multiple motion model estimation. • Lifting developmental constraints is regarded • Runs in an online and incremental as the force to drive the robot to develop. fashion. • Experimental results show the robot owns human-like developmental features.

81 IEEE ROBIO 2012 Conference Digest

Th2-4: Multi-Robot Systems II Session Chairs: Wei Pan and Lizhen Wang

Room: Jade 8, 13:30-15:00, Thursday, December 13, 2012

Th2-4(1) 13:30-13:45 Th2-4(2) 13:45-14:00

Dynamic Bandwidth Management for Cooperative Transportation Control in Consideration of not only Teleoperation of Collaborative Robots Internal Force but also External Force Applied to ``MRWheel'' Koshi Kashiwazaki, Kazuhiro Kosuge and Yasuhisa Hirata 1 2 1 2 Chadi Mansour , Elie Shammas , Imad H. Elhajj and Daniel Asmar Department of Bioengineering and Robotics, Tohoku University, Japan 1 Electrical and Computer Engineering Department Yusuke Sugahara 2 Mechanical Engineering Department School of Science and Engineering, Kokushikan University, Japan American University of Beirut, Lebanon Takashi Kanbayashi, Koki Suzuki, Kazunori Murakami and Kenichi Nakamura IHI Transport Machinery Co., Ltd., Japan • Introduced a bandwidth management • We proposed a car transportation system optimization algorithm based on task Transported Car A figure here referred to as ``iCART II’’ shown in the figure. conditions and quality of collaboration for swarm of networked robots • This paper proposes a cooperative transportation control in consideration of • Algorithm validated by simulating the behavior not only internal force but also external force. of a swarm of robots teleoperated using a haptic device while providing visual and haptic • We compare the proposed control algorithm with the conventional one, and the results of MRWheel feedback For figure caption (optional), car transportation experiments confirm the (Mobile Robot for a Wheel) • Results demonstrated improvements in the use arial narrow 20pt font validity of the proposed one. Fig. Appearance of iCART II. network and task performance

Th2-4(3) 14:00-14:15 Th2-4(4) 14:15-14:30

Estimation of Relative Positions and Discrete Analysis of Object Clustering by Orientations on a Car Transportation System Distributed Robots Grasping two Drive Wheels Yuichiro Sueoka,Takuto Kita, Masato Ishikawa, 1 1 1 2 N. Yonezawa , K. Kosuge , Y. Hirata , Y. Sugahara , Yasuhiro Sugimoto, Koichi Osuka T. Kanbayashi3, K. Suzuki3, K. Murakami3 and K. Nakamura3 1Department of Bioengineering and Robotics, Tohoku University, Japan Dept. of Mech. Eng., Osaka Univ, Japan 2School of Science and Engineering, Kokushikan University, Japan 3 IHI Transport Machinery Co., Ltd., Japan • We discuss object clustering by distributed • This presentation is about a car autonomous robots, in the light of space- transportation system consisting of discretization approach. two mobile robots. • We introduce two types of local rule, sensor • This presentation proposes an avoiding rule and push & turn rule. estimation algorithm of some positional relationships between the • We analyze the transient/steady- state robots and the car. behaviors and density of objects and robots.. • The algorithm uses force sensor Object clustering by distributed robots systems and encoders embedded on Overview and geometrical relationships with spatial discretized system the robots. of a car transportation system

Th2-4(5) 14:30-14:45 Th2-4(6) 14:45-15:00 Part Dispatching Rule-Based Multi-robot Distributed Formation Control for a Cooperative Coordination in Pick-and-Place Task Multi-agent System Using Potential Function Yanjiang Huang and Jun Ota and Extremum Seeking Algorithm Research into Artifacts, Center for Engineering, The University of Tokyo, Japan Heng Li,Jun Peng,Jianming Xiao,Feng Zhou and Weirong Liu Ryosuke Chiba School of Information Science&Engineering, Central South University, China Faculty of System Design, Tokyo Metropolitan University, Japan Jing Wang Tamio Arai Computer Engineering Program, Bethune-Cookman University, USA Department of Mechanical Engineering, The University of Shibaura Institute of Technology, Japan • A distributed formation scheme integrating Tsuyoshi Ueyama potential function and extremum seeking A figure here DENSO WAVE INCORPORATED, Japan algorithm is proposed. Robot arm • Part dispatching rules are used to coordinate Camera • A comprehensive performance index related to multi-robot system. the environment is proposed. • Greedy randomized adaptive search procedure • The desired separation distance is obtained (GRASP) is used to search for appropriate Picking tool dynamically by maximizing the performance combination of part dispatching rules. Conveyor with extremum seeking algorithm. The integrating scheme • GRASP can converge to good solution with low Parts computational time. A multi-robot conveyor system

82 IEEE ROBIO 2012 Conference Digest

Th2-5: Rehabilitation Robots III Session Chairs: Yuji Okada and Jun Peng

Room: Jade 5, 13:30-15:00, Thursday, December 13, 2012

Th2-5(1) 13:30-13:45 Th2-5(2) 13:45-14:00

Parameterized Inverted and Double Pendulum An Electrical Muscle Simulator Based on Model for Controlling Lower-limb Active Orthosis Functional Electrical Stimulation Juan Carlos Arevalo, Daniel Sanz-Merodio, Haibin Wang Shenzhen Institutes of Advanced Technology , Manuel Cestari and Elena Garcia Guangzhou Institute of Advanced Technology, CAS, China Guan Guan, Qing He, Dewen Zeng, Bin Leng, Hongwei Xu, Weiming Zheng Field and Service Robotics Group, Centre for Automation and Robotics Guangzhou Institute of Advanced Technology, CAS,China (Spanish Research Council), Spain • The paper describes a trajectory generation • The basic principal of electrical sim ulation is described. method for an active lower-limb orthosis. A figure here • The orthosis will be used by a quadriplegic girl • The design of an electrical muscle and includes a walking-frame for stability. simulator is presented, including b oth hardware and software. • The method divides human gait in three simple models. • Some experimental results to verif y the feasibility and performance of • The trajectories are generated by the proposed simulator. parameterizing the position outputs obtained from the models. • A brief conclusion and future work are given.

Th2-5(3) 14:00-14:15 Th2-5(4) 14:15-14:30 Basic Experiments on Collision of Mechanical Design of a Magnetic Resonance Sharp Mechanical Hazards against Eye for Compatible Device Used for Finger Estimation of Injury Severity Rehabilitation and Brain Imagine Studying Soichiro Ito, Yoji Yamada, Takamasa Hattori, ZhenJin Tang, Shigeki Sugano and Hiroyasu Iwata Shogo Okamoto, and Susumu Hara Department of Modern Mechanical Engineering, Graduate School of Engineering, Nagoya University, Japan Waseda University, Tokyo, Japan.

• We deal with a situation in which a ● MRI Compatibility, Portability robot’s sharp end effector attacks a ● Adjustable to a variety of hand sizes. human eye. Both fitting to various lengths of • We conduct eye collision experiments phalanges and different gaps between taking into consideration human Porcine eyeball fingers. avoidance/mitigation actions. ● Avoiding interference between the hand exoskeleton and the finger. • It is confirmed that differences in End effector ● Two DOF motion per finger (MCP and Prototype of MRI severity result from the changes in the compatible finger PIP) and one DOF motion in Thumb. collision conditions. rehabilitation robot ● User safety.

Th2-5(5) 14:30-14:45 Motion Support of Upper Extremity with Agonist alone under Negative Admittance Control Yuji Okada*, Chi Zhu*, Simazu Shota*, Toshikatsu Suzuki*, Masataka Yoshioka*, Yuichiro Yoshikawa*, Yuling Yan**, Haoyong Yu***, and Feng Dung**** *Maebashi Institute of Technology, Japan. ** Santa Clara University, USA ***National University of Singapore, Singapore. **** Nankai University, China ・A negative admittance control is originally developed to support the movement of human’s upper extremity only by agonist’s EMG signals. ・ In the control, two virtual parameters, virtual damping coefficient and virtual moment of inertia, are negative. ・The experimental results show that the performance of the negative admittance control can be arbitrarily adjusted by control parameters and user’s “load feeling” can be arbitrarily reduced.

83 IEEE ROBIO 2012 Conference Digest

Th2-6: Robot Control II Session Chairs: Yongsheng Ou and Zhicheng Qiu

Room: Jade 9, 13:30-15:00, Thursday, December 13, 2012

Th2-6(1) 13:30-13:45 Th2-6(2) 13:45-14:00 The analysis and synthesis of PID controller Adaptive Fuzzy Tracking Control of a Human Lower based on closed loop response characteristics Limb With an Exoskeleton Zhu Zhiqiang State Key Laboratory of Robotics, Shenyang Institute of Automation Yongsheng Ou, Zhijun Li, Guanglin Li, and Chun-Yi Su Graduate University of Chinese Academy of Sciences, China He Yuqing and Qi Juntong and Han Jianda and Wang Tianran develop a lower limb State Key Laboratory of Robotics, Shenyang Institute of Automation, China exoskeleton to produce a knee position trajectory • New PID parameters are proposed to Lyapunov-based adaptive fuzzy control express the magnitude and phase is developed to control properties of PID controller clearly and the human quadriceps femoris muscle regularly. undergoing nonisometric • Illustrative graphical analysis of systems contractions. with time delay needs no approximation. The experiments illustrate the ability of • PID tuning rules that don’t need plant the exoskeleton model. The Nyquist curves of the PID to enable the leg shank to track single and control systems with different Kd multiple period Trajectories.

Th2-6(3) 14:00-14:15 Th2-6(4) 14:15-14:30 A Direct Visual Servo Scheme Based on Fuzzy Moving Fast Terminal Sliding Mode Simplified Interaction Matrix for High-Speed Control for Robotic Manipulators Manipulation H.J. Shi, L.F. Qian, Y.D. Xu and L.M. Chen Shouren Huang, Yuji Yamakawa, Taku Senoo and Masatoshi School of Mechanical Engineering, Nanjing University of Science and Ishikawa Technology, China Graduate School of Information Science & Technology, Univ. of Tokyo, Japan

• Nonsingular terminal sliding mode control with • Simplified interaction matrices for image-based fuzzy tuning. visual servoing with high-speed camera were • A class of nonlinear systems in the presence of proposed uncertainties and external disturbance. • Direct visual servo approach for high-speed • The sliding surface can rotate and bend in the manipulation was realized phase space. • Comparison simulations, experiments for high- • The tracking behavior can be improved, and speed ball tracking, and for peg-and-hole chattering can also be reduced. the desired and actual alignment were conducted trajectories (a)High-speed ball tracking (b) Peg-and-hole alignment

Th2-6(5) 14:30-14:45 Th2-6(6) 14:45-15:00

Pneumatic Drive Active Vibration Control for Integrated Rotary Compliant Joint and Its Flexible Manipulator Using an Adaptive Impedance-based Controller for Single-Joint Interactive PD Controller Pressing Massage Robot Yuancan Huang, Chongjian Ran, Jian Li and Guodong Li Zhicheng Qiu and Zhili Zhao Institute of Intelligent Robot, Beijing Institute of Technology, China Mechanical and Automotive Engineering, South China University of Technology, Guangzhou, China • A compact, integrated rotary • A pneumatic drive system is used as the compliant joint was designed and A figure here actuator for active vibration control of fabricated. flexible manipulator. • An impedance-based controller with • An adaptive interactive PD controller is joint torque control was proposed. applied to positioning and vibration control. • The effectiveness of the designed • Results from simulations and experiments integrated rotary compliant joint was showed the feasibility of the actuator and validated by a simple single-joint the effectiveness of the control strategy. pressing massage robot. Test-bed of integrated compliant joint

84 IEEE ROBIO 2012 Conference Digest

Th2-7: Sensors I Session Chairs: Haoting Liu and Jinping Hou

Room: Jade 7, 13:30-15:00, Thursday, December 13, 2012

Th2-7(1) 13:30-13:45 Th2-7(2) 13:45-14:00

Defect Segmentation of Fiber Splicing on an Numerical model of a novel tactile sensor System Using GMM and Graph Cut based on finite element analysis Haoting LIU, Wei WANG, Xinfeng LI Xin Sun1, Xuekun Zhuang1,2, Hongqing Pan1, Yaoxiong Wang1, Beijing Aerospace Times Optical-electronic Technology Co. Ltd., China Yubing Wang1, Xubin Liang3, Quanjun Song1 and Feng Shuang1,2 Fan LI 1. Institute of Intelligent Machines, Chinese Academy of Sciences, China Science and Technology on Human Factors Engineering Laboratory, Astronaut 2. Department of Automation, University of Science and Technology of China Research & Training Center of China, China 3. Engineering Practice and Training Center, Guangxi University,China • A novel technique is presented to solve the • A preliminary numerical model of a three- region segmentation problem of the fiber dimensional tactile sensor based on FEA (finite splicing process on our industrial robot system. element analysis) is proposed. • Both images of normal mode and those of • It builds the mapping between the resistances defect mode are utilized as prior information. and the deformation of the tactile sensor. • The Gaussian Mixture Model is used to restrain • A decoupling algorithm based on the RBF the high light region of spliced fiber. (radical basis function) network is applied to • The Graph Cut Model is employed to segment the numerical model, and the results show that the numerical model of the tactile sensor the defect region and analyze their features. this algorithm can perform on the low- dimensional numerical model quickly.

Th2-7(3) 14:00-14:15 Th2-7(4) 14:15-14:30

Developing tactile sensors for a soft-bodied A Model-based Human Walking Speed Estimation robot Using Body Acceleration Data Jinping Hou, and George Jeronimidis Jwu-Sheng Hu, Kuan-Chun Sun, Chi-Yuan Cheng Centre for Biomimetic, University of Reading, UK Department of Electrical and Control Engineering, Richard HC Bonser National Chiao-Tung University, Taiwan, R.O. C. School of Engineering and Design, Brunel University, UK •We propose a novel virtual inverted pendulum model. • Developing and testing of tactile sensors for a This model not only keeps the important characteristic in robot inspired by octopus biped rolling-foot model, but also makes the speed • QTC sensors were fabricated and tested estimation feasible using human body acceleration. Soft sensors developed using silicone rubber • •In addition, this model also includes the effect of rotation and electrolycra have adjustable sensitivity, size and shape, and can extend over 100% in of waist within a walking cycle that improves the length estimation accuracy. • A whole arm skin prototype with embedded •Experimental results on a group of humans show a sensors are presented 1.22% error mean and 2.78% deviation, which is far better than other known studies.

Th2-7(5) 14:30-14:45 Th2-7(6) 14:45-15:00

Multiple Emitter Localization Using Range Only Twin Low-cost Infrared Range Finders for Measurements Considering Geometrical Constraints Detecting Obstacles Using in Samitha Ekanayake and Pubudu Pathirana Mobile Platforms School of Engineering, Deakin University, Australia Saman Kumpakeaw College of Industrial Technology, Terry Caelli King Mongkut's University of Technology North Bangkok, Thailand National ICT Australia, Austrlia • Constrained optimization approach to • Elimination of blind spots for range detections improve the accuracy of a Time-of-Arrival due to sensor limitations. (ToA) based multiple target localization • Measurement compensation using two low- system. cost infrared sensors with different ranges. Uses local distance measurements between • • The sensors provide distance measurement the targets/emitters as the geometric capacity from 20-500 cm. (compared to constraint. Hokuyo URG-04LX : 2-400 cm) • Applications in the ToA based human motion Overview of the constrained capture systems • Preference to use in in optimization algorithm indoor environments for SLAM and navigations.

85 IEEE ROBIO 2012 Conference Digest

Th3-1: Modular Robot Systems Session Chairs: Michael Goeller and Takuto Kita

Room: Jade 1, 15:20-16:50, Thursday, December 13, 2012

Th3-1(1) 15:20-15:35 Th3-1(2) 15:35-15:50

Conceptual Study of a Class of Hyper-Redundant Geometric Modelling of Multisection Bionic Robot Manipulator: Experimental Validations on Abdelhakim Chibani and Rochdi Merzouki RobotinoXT LAGIS, Polytech’ Lille, France C. Escande1&2 , R. Merzouki1, P. M. Pathak3, and V. Coelen1 Chawki Mahfoudi 1 LAGIS UMR CNRS 8219, Polytech'Lille, Villeneuve d'Ascq, France Faculty of Technology , Univ. Larbi Ben M’Hidi, Algeria 2 Festo-Didactic, Denkendorf, Germany Taha Chettibi 3 Mechanical & Industrial Engineering Department, IIT, Roorkee, India EMP, Algiers, Algeria • Kinematic analysis of HHRR manipulator based • Synthesis of Forward Geometric Model of on the theory of parallel robot; Bio-inspired Flexible Manipulator and • Inverse geometric model generated by Experimental Validation for Multisection approximating robot backbone curve using • Synthesis of Inverse Geometric Model for cubic spline; One Section and Experimental Validation • Simulation results; for One Section • Future work: development of an effective • Future Improvements on the Geometric RobotinoXT control process in normal and faulty operations. Model of RobotinoXT

Th3-1(3) 15:50-16:05 Th3-1(4) 16:05-16:20 Modular Robots for On-Orbit Satellite Servicing Performance Analysis and Optimization Michael Goeller, Jan Oberlaender, Klaus Uhl, Arne Roennau, and Rüdiger Dillmann FZI - Research Center for Information Technology, 76131 Karlsruhe, Germany

of Serial Robots [goeller, oberlaender, pfotzer, roennau, dillmann]@fzi.de Wenqiang Wu1, Yisheng Guan1, Huaizhu Li1, Haifei Zhu1, and Hong Zhang1, 2 Today satellites are mostly monolithic systems and offer no options to facilitate servicing or 1 Biomimetic and Intelligent Robotics Lab (BIRL) maintenance. In this work, supported by the German Space Agency DLR, we propose a School of Mechanical and Automotive Engineering modular concept for satellites based on South China University of Technology standardized building blocks. 2 Department of Computing Science, University of Alberta, Canada These building blocks could be replaced in orbit requiring significantly less manipulation skills than ordinary repair missions would need.  A comprehensive performance criteria for serial robots is proposed The benefits would be: • Prolonged lifetime of satellites  An optimization algorithm based GO • Reduced costs • Less space debris and normalization method is presented • Faster development of satellites

 Performance analysis and dimension In this paper we will present solutions for major optimization is conducted with the challenges which are strongly inspired by earthbound heterogeneous robotic systems: biped climbing robot Climbot. • Standardized interfaces for data, electrical current, and mechanical coupling  The CAD model of Climbot and its Dimension optimization of a 6-DoF • High-bandwidth data-bus concept manipulator is also performed reciprocal condition number • Software framework for dynamic systems © FZI Forschungszentrum Informatik

Th3-1(5) 16:20-16:35 Th3-1(6) 16:35-16:50 On Discrete-valued Modeling of Parameterized Space Conditions for the Definition Nonholonomic Mobile Robot Systems of Locomotion Modes in Modular Snake Robots Takuto Kita, Masato Ishikawa and Koichi Osuka Kamilo Melo, Monica Hernandez and Daniel Gonzalez Dept. of Mech. Eng., Osaka Univ, Japan KM-ROBOTA Research Group, KM-ROBOTA S.A.S., Colombia • Instead of the special Euclidean space SE(2), we suppose the hexagonal cellular space as the field • Mathematical representation of the of planar locomotion. parameterized space and its mapping to configuration, work and shape spaces are • We propose models of discrete- valued nonholonomic systems such described formally. as single cart, trailer system and • Constrains of the parameterized control space snake robot. for modular snake robot are shown. • We examine the reachabillity of • Formal mathematical definition of a locomotion system states in discrete settings and Stepwise reachability set of mode for modular snake robots is presented. Parameterized Gaits discuss the controllability. the single trailer system (step7) • Locomotion mode usage is presented (gait design representation experiments carried out).

86 IEEE ROBIO 2012 Conference Digest

Th3-2: Motion Planning III Session Chairs: Corrado Guarino Lo Bianco and Dewen Zeng

Room: Jade 2, 15:20-16:50, Thursday, December 13, 2012

Th3-2(1) 15:20-15:35 Th3-2(2) 15:35-15:50

An improved ant colony optimization algorithm Experimental validation of a time scaling based on dynamically adjusting ant number algorithm for robotics systems Corrado Guarino Lo Bianco and Fabio Ghilardelli Dewen Zeng,Qing He,Bin Leng Dipartimento di Ing. dell'Informazione, University of Parma, Italy Intelligent control research center, Guangzhou Institutes of Advanced Daniel Kubus Technology, Chinese Academy of Sciences, China Institut für Robotik und Prozessinformatik, TU Braunschweig, Germany

• Improved ant colony optimization algorithm by dynamically adjusting ant • Time scaling system to avoid the occurrence of number. kinematic/dynamic saturations. • update the total ant number randomly • The algorithm is tested on a four links, closed- or fixedly in algorithm iterative process chain system. • It is easier to find the optimal solution, • The scaling system ensures the fulfillment of and its optimization ability is stronger. the voltage, velocity and acceleration constraints.

Th3-2(3) 15:50-16:05 Th3-2(4) 16:05-16:20 Euclidean Distance and Workspace Region Based An Electromyography-driven Central Pattern Control Algorithm for Collision Avoidance in a Generator Model for Robotic Control Application Laparoscopic Surgical Robot: MU-LapaRobot KuiXiang, Shuxiang Guo, Zhibin Song Wuhan University of Technology, China; Kagawa University, Takamatsu, Japan Jitendra Yadav and Jackrit Suthakorn • A new CPG model driven by surface Center for Biomedical and Robotics Technology (BART LAB); electromyography (sEMG) was Department of Biomedical Engineering, Mahidol University, Thailand presented to manipulate robots. • The MU-LapaRobot is designed to hold tools • The CPG model was designed based on an echo state network (ESN). during minimally invasive surgery. A figure here • The CPG model could learn and • An algorithm for controlling the collision generalize the frequency transition avoidance of tools during Minimally Invasive pattern. Surgery. • The flexion and extension motion • Control of speed of tools around the operating and the corresponding sEMG were regions according to the Euclidean distances. sampled to train a CPG model for • Movement based according to different rehabilitation exoskeleton control. Tools and different layers regions defined around the operating regions.

Th3-2(5) 16:20-16:35 Th3-2(6) 16:35-16:50

SINGLE LOOP AND DOUBLE LOOP BALANCING Cartesian robot motion embedded on Logistics CONTROL OF AU SELF-BALANCING BICYCLE Process, screws approach (AUSB) Mario Arbulú IEEE Member, German Ortiz, Luis M. Beltrán, Narong Aphiratsakun and Kittiphan Techakittiroj Claudia Garzón Faculty of Mechatronics Engineering, Informatics Program, Universidad de La Sabana, Colombia Assumption University, Thailand Leonardo Gonzalez Industrial Program, Universidad de La Sabana, Colombia • A Comparison between existing Single loop • Novel analytical kinematics model, for cartesian (sensor: gyro) with an additional Double loop robots, with screws and 6D vectors instead of (Sensors: gyro and encoder). local frames in 3D space. • Gyro loop is used to keep bicycle in the • Jacobian computation without partial derivate, it balancing position, but not controlling flywheel depends of twits and joint position. to be in the horizontal position. • That is, the jacobian is a set of 6D vectors; and • Encoder loop is used to keep flywheel to be in it will be used in dynamic model on generalized the horizontal position, hence the balancing inertial matrix performance is improved. Cartesian Robot model • Smooth joint motion without singularities, because local joint frames are avoided.

87 IEEE ROBIO 2012 Conference Digest

Th3-3: Vision/Image Processing IV Session Chairs: Zhidong Wang and Chris Smith

Room: Jade 3, 15:20-16:50, Thursday, December 13, 2012

Th3-3(1) 15:20-15:35 Th3-3(2) 15:35-15:50 A Two-Stage Strategy to Introduce Spectral Matching Image-based Tree Pruning into Recognition of Occluded Objects Wu Jiayun Wei Liu and Nanning Zheng Mechatronics and Control Lab, National University of Singapore, Sigapore AI&R Institute, Xi’an Jiaotong University, China Chen Xiao George Kantor and Fernando De la Torre Research Institute of Computer Application, China Academy of Engineering Robotics Institute, Carnegie Mellon University, USA Physics • Tree pruning with a high-resolution and a • Recognition of partially visible objects in a Kinect camera mounted on a mobile robot. A figure here scene A figure here • Fusing spatial and temporal information in • This local to global nature of occlusion depth images. recognition leads us to spectral matching • Capture and segment small branches. technique; • Quantitative estimation of the angles and • a top- down procedure is introduced into length for each branch. spectral matching for the recognition of Feature points associated with For figure caption (optional), occluded objects. objects of interest use arial narrow 20pt font

Th3-3(3) 15:50-16:05 Th3-3(4) 16:05-16:20

Vanishing Point Detection Using Local Dominant Object Extraction from Stereo Vision Orientation Signature for Omnidirectional using Continuity of Disparity Map Vision-Based Robots Kotaro Oshida Physics, Tokyo Institute of Technology, Japan Jun-Yu Yang, Feng-Li Lian, and Jiun-Jau Lai Department of Electrical Engineering, National Taiwan University, Taipei Taiwan Keiji Saneyoshi Radiation Research and Management Center, Tokyo Institute of Technology, Japan • A novel method based on the • Disparity distribution is discontinuous omnidirectional vision to tackle the tasks A figure here at borders between objects. related to vanishing point detection is • Base image is separated into superpixels. proposed. Superpixels are approximated by planes • The orientation information in the panoramic in disparity map. image is first computed by the Sobel edge • A superpixel is connected to adjacent detection. superpixels, if their planes are continuous. • The orientation information is next analyzed • The processed superpixels show objects. by local dominant orientation signature.

Th3-3(5) 16:20-16:35 Th3-3(6) 16:35-16:50

Unscented Blind Image De-blurring Using Supporting System for Self Training of Bed-Making Using Camera with Inertial Measurement Unit Image Processing with Color and Distance Information Ayanori Nagata and Zhifeng Huang Dept. of Precision Engineering, the University of Tokyo, Japan Chin-Yuan Tseng, Jian-An Chen and Jwu-Sheng Hu Masako Knai-Pak, Jukai Maeda, Yasuko Kitajima, Mitsuhiro Nakamura and Kyoko Aida Department of Electrical and Control Engineering, National Chiao-Tung Faculty of Nursing, Tokyo Ariake University of Medical and Health Science, Japan University, Hsinchu, Taiwan, ROC Noriaki Kuwahara Dept. of Advanced Fibro-Science, Kyoto Institute of Technology, Japan Taiki Oagata and Jun Ota Research into Artifacts Center for Engineering, the University of Tokyo, Japan • We proposes a sensor fusion model for blind image de-blurring using inertial measurement unit. • Developed a support system that measures • The model attempts to observe the camera motion, estimate the point spread and evaluates the nursing trainees function and de-convolute the image simultaneously. movement of bed-making. A figure here • An iterative estimation procedure using MAP-EM algorithms and Unscented • Determined evaluation points for bed- Kalman Filter are proposed. making and constructed measurement system with Kinects. Schematic image of a • By the sensor fusion formulation, we are able to cope with the effect of • Evaluated the accuracy of the system by supporting system to assist unknown bias and drift of the IMU sensor. nursing students in learning comparing the evaluation of nursing certain nursing skills teachers.

88 IEEE ROBIO 2012 Conference Digest

Th3-4: Multi-Robot Systems III Session Chairs: Jun Peng and Hongliang Ren

Room: Jade 8, 15:20-16:50, Thursday, December 13, 2012

Th3-4(1) 15:20-15:35 Th3-4(2) 15:35-15:50

Transportation of a large object by small mobile robots using hand carts

Takuya Sakuyama, Jorge Figueroa, Yuta miyazaki and Jun Ota Research into Artifacts, Center for Engineering (RACE), The University of Tokyo

• Methodology for transportation of a large object by mobile robots using small hand carts. • An algorithm that decides how to insert the hand carts. • We designed an end-effector in order to realize our method in a real world environment.

Th3-4(3) 15:50-16:05 Th3-4(4) 16:05-16:20

Analysis of Implicit Control Structure in Object Hearing Based Relative Localization for Mobile Clustering Phenomena Robots in Outdoor Environments Yuichiro Sueoka, Yasuhiro Sugimoto, Masato Ishikawa, Koichi Osuka Jing Guo, Qing-Hao Meng, Member, IEEE, Yu-Xiu Wu, Wei-Xing Dept. of Mech. Eng., Osaka Univ, Japan Yang, Ming Zeng, Member, IEEE, and Wei Li, Member, IEEE Akio Ishiguro School of Electrical Engineering and Automation, Tianjin University, China Research Institute of Electrical Communication, Tohoku Univ. Japan • Some simple robots named Swiss Robot, • A hearing based relative localization method Aggregator Robot, Coronoc Robot, are for mobile robots in outdoor environments is A figure here developed. presented. • These robots show object clustering behaviors. Coronoc Robot • A device including a buzzer and an array of our • From the systematic analysis, we can see microphones is designed. common Implicit Control Law. • The accuracy of the relative localization is • We demonstrate the correspondence between characterized through some tests. the Explicit Control Law and its clustering Swiss Robot • The usefulness of the proposed method is ability. demonstrated through a formation experiment Coronoc Robot and Swiss Robot

Th3-4(5) 16:20-16:35 Towards Cloud Robotics System: A Case Study of Online Co-localization for Fair Resource Competence Lujia Wang1 and Ming Liu2 and Max Q.-H Meng1 1Department of Electronic Engineering, The Chinese University of Hong Kong 2Autonomous Systems Lab, ETH Zurich, Switzerland

• A generic framework, which is a twisted socket-based server-client platform with ROS as the data retrieval, is introduced for real-time co-localization. • The multi-robot autonomous negotiation (MRAN) module is proposed which relieves the competition among robots. • A real-time scenario, enabling several poor-equipped robots to retrieve location from a updated map, is validated in an indoor environment. 3D point cloud and map for multi- robot co-localization

89 IEEE ROBIO 2012 Conference Digest

Th3-5: Search Rescue Robots Session Chairs: Daniel Asmar and Changlong Ye

Room: Jade 5, 15:20-16:50, Thursday, December 13, 2012

Th3-5(1) 15:20-15:35 Th3-5(2) 15:35-15:50

Kinematic Analysis of an Active Angular Swivel Design, Implementation and Experimental Tests Steering Mechanism for Robotic Mobile Bases of A New Generation of Antarctic Rover

Elie Shammas and Daniel Asmar Cheng Chen and Chunguang BuYuqing He and Jianda Han Mechanical Engineering Department, American University of Beirut State Key Laboratory of Robotics, Shenyang Institute of Automation, Beirut, Lebanon University of Chinese Academy of Sciences, China • Novel active steering mechanism known as angular swivel steering. • This paper introduces the hardware and software design and implementation of • New mechanism could be potentially the third generation of Antarctic rover we exploited to negotiate rugged terrain. designed. The preliminary experimental • Experimental analysis shows the results during the 28th Chinese National possibility of steering angles up to 90 Novel active steering mechanism Antarctic Research Expedition , suggest degrees (rotate in place) that the design and implementation of the • During steering one wheel is located above the platform, a fact that hardware and software system can would allow the robot to negotiate relatively high obstacles. ensure efficiency and reliability of the rover working on the Antarctic.

Th3-5(3) 15:50-16:05 Th3-5(4) 16:05-16:20 Development of a Variable Parallelogram The Design of a Throwable Two-Wheeled Tracked Mobile Robot Reconnaissance Robot Hengrui Gao, Shusheng Bi and Rong Zhang Changlong Ye, Guangming Lv, Shugen Ma and Huichao Ni School of Mechanical Engineering and Automation, Beihang University, China College of Electromechanical Engineering,Shenyang Aerospace University, Shichuan Tang China Beijing Municipal Institute of Labour Protection, China Robotics, Ritsumeikan University, Japan

• Design of a throwable two-wheeled robot which • This VPTMR robot consists of two variable can withstand landing impact from 10m height. parallelogram structures, which are composed A figure here • It is based on co-centric shaft structure and of one main tracked arm, two lower tracked bionic outer shell. arms and a chasis. • Simulation of non-linear drop impact. • The VPTMR robot to move in wheeled, tracked • Optimize of rubber wheels by LS-DYNA. and legged mode that makes the robot to adapt to all rugged environments. BHTBOT-3 • The prototype(VPTMR) is developed to verify Variable parallelogram tracked its performance on environmental adaptability, mobile robot obstacle crossing ability and stability.

Th3-5(5) 16:20-16:35 Th3-5(6) 16:35-16:50

Development of an Underactuated Robot Gripper Multi-Robot Repeated Boundary Coverage Capable of Retracting Motion Under Uncertainty

Hiroyuki Nishimura, Atsushi Kakogawa Pooyan Fazli and Alan K. Mackworth Department of Computer Science and Shugen Ma University of British Columbia Department of Robotics, Ritsumeikan University, Japan Vancouver, BC, Canada V6T 1Z4

• We present a new underactuated robot gripper, • We address the problem of repeated coverage which can grip various objects with retracting by a team of robots of the boundaries of a target motion. area and the structures inside it. • Events may occur on any parts of the • The gripping and retracting motions are given boundaries and may have different importance by a single motor. weights. • A rotational resistance mechanism regulates • The goal is to maximize the reward by detecting the maximum number of events, the distribution of the motor power. weighted by their importance, in minimum time. • To verify the validity, the experiments of the Overview of the gripper fabricated prototype are conducted.

90 IEEE ROBIO 2012 Conference Digest

Th3-6: Robot Control III Session Chairs: Emmanuel Dean-Leon and Seiji Sugiyama

Room: Jade 9, 15:20-16:50, Thursday, December 13, 2012

Th3-6(1) 15:20-15:35 Th3-6(2) 15:35-15:50

Adaptive Harmonic Compensation of Shunt Active Power Filter Using Sliding Mode Controller Juntao Fei, Shenglei Zhang and Jian Zhou User Friendly Matlab-Toolbox for Symbolic Robot College of Computer and Information, Hohai University, China Dynamic Modeling used for Control Design Emmanuel Dean-Leon † , Suraj Nair †, Alois Knoll ‡, Member, IEEE † Cyber-Physical Systems, fortiss - An-Institut der Technischen Universität München. ‡Technische Universität München, Fakultät für Informatik. † {dean, nair}@fortiss.org, ‡[email protected] • An adaptive sliding mode controller is proposed to implement the harmonic compensation for In this paper a new Robot Modeling/Simulation Toolbox for Matlab the single-phase shunt APF. A figure here is presented. It can compute the kinematic and dynamic equations of a • Tracking error of harmonic and APF current as serial robot in closed-form.

the control input, and tracking error of reference The primary contribution is the model and APF output as the control objects of automation and simplification of the robot modeling process which is adaptive sliding mode. important for correct robot design and control. • Simulation results demonstrate the satisfactory harmonic eliminating performance , rapid compensation ability of the proposed controller.

Th3-6(3) 15:50-16:05 Th3-6(4) 16:05-16:20

Stabilization of an Unmanned Aerial Vehicle Guide Robot with Personal Identification Method Using Real-Time Embedded Motion Estimation Using Dress Color Information via KINECT Matthieu Claybrough Seiji Sugiyama*, Kouhei Baba*, and Tsuneo Yoshikawa* École Polytechnique, France *Department of Human and Computer Intelligence, François Defaÿ College of Information Science and Engineering, DMIA, ISAE, France Ritsumeikan University, Japan

• Efficient tracking algorithm for • A Simple Personal Identification method using estimating global motion with Dress Color Information for guide robots has been proposed. minimal image processing • Color information is only calculated at narrow areas around user’s joint positions obtained • Real-time embedded system for via KINECT. autonomous stabilization • Although the robot loses the user temporarily, it can find the user properly and promptly. • Vision-based controller for position • Experiments that show the accuracy of our Our mobile robot can identify the hold and velocity control method and an example of guidance are specific user via KINECT, and can guide him to the destination presented. using the line trace method

Th3-6(5) 16:20-16:35 Th3-6(6) 16:35-16:50

An Open Control System Architecture with an Proposal for An IR System to Support On-line Velocity Filter for Industrial Robots Automatic Control for A Personal Mobility Vehicle Zheng Wang and Yunjiang Lou Harbin Institute of Technology Shenzhen Gratuate School, China Taro Fujikawa and Shuro Nakajima Yue Liu Dept. of Advanced Robotics, Chiba Institute of Technology, Japan Googol Technology (Shenzhen) Ltd, China Zexiang Li Hong Kong University of Science and Technology, China • We propose a supporting system for automatic • Open control system architecture for general control of personal mobility vehicles (PMVs) to industrial robot. pass through a narrow space in indoor IR marker • Both hardware and software of the robot environments. control system is discussed in detail. Narrow space • This system consists of IR markers located on θ • An on-line velocity filter with the function of an entrance of a narrow space and an IR sensor l look-ahead is proposed. to mount on a PMV. • An IR sensor calculates relative distance and • Integrate this algorithm to the control system PMV and validate the performance by experiments. angle of approach to pass through on an IR marker. IR sensor

91 IEEE ROBIO 2012 Conference Digest

Th3-7: Sensors II Session Chairs: Jian Guo and Zhuohua Lin

Room: Jade 7, 15:20-16:50, Thursday, December 13, 2012

Th3-7(1) 15:20-15:35 Th3-7(2) 15:35-15:50 Performance Evaluation of the Wireless Inertial Development of Force Sensing Systems for a Measurement Unit WB-4 with Magnetic Field Calibration Novel Robotic Catheter System Z. Lin1, M. Zecca2, S. Sessa2, L. Bartolomeo1, Jian Guo *1, Shuxiang Guo *1,*2, Nan Xiao*2 and Yunliang Wang *1 H. Ishii3, and A. Takanishi4 1 Global Robot Academia, Waseda University, Japan 2 *1 School of Electrical Engineering, TianJin University of Technology, China School of Creative Science and Engineering, Waseda University, Japan 3 *2Faculty of Engineering, Kagawa University, Japan Research Institute for Science and Engineering, Waseda University, Japan 4Department of Modern Mechanical Engineering, Waseda University, Japan • This paper presents two kinds of force • Development of a wireless ultra-miniaturized sensing systems for a novel robotic Inertial Measurement Unit WB-4 IMU catheter system. • Methodology proposed for the IMU • We did the experiments to evaluate the performance evaluation with magnetic field developed two kinds of force sensing calibration function systems. • Experimental validation of the IMU • The experimental results indicated that performance with the VICON optical system Experimental setup for the the developed two kinds of force sensing The prototype of one kind of force • Experimental results show the WB-4 IMU has performance evaluation of WB-4 sensing system systems are effective. high dynamic performance after magnetic IMU with VICON system calibration proposed in the paper

Th3-7(3) 15:50-16:05 Th3-7(4) 16:05-16:20

Optimal Sensor Placement in Range Based Localization with Ghost Elimination of Emitters Localization for Linear Arrays Via Time-of-Arrival Measurements

Sanvidha Herath, Pubudu N. Pathirana and Gareth Williams Sanvidha Herath, Pubudu N. Pathirana and Nicholas Boer School of Engineering, Deakin University, Australia School of Engineering, Deakin University, Australia

• Investigates the linear separation requirements • Analysis on the unique localization of for range sensors in order to achieve the synchronized multiple emitters using Time-of- optimal performance in estimating the position Arrival measurements subjected to the data- of a target from multiple and typically noisy association problem. sensor measurements. • The necessary fundamental requirements to • The sensor-target geometry is analysed in solve the so-called ghost node problem terms of the Cramer-Rao inequality and the associated with sensor arrays are examined. corresponding Fisher information matrix, in • Derivation of a measurement bound for ideal order to characterize localization performance situations and the underlying concepts are with respect to the linear spacial distribution. illustrated via simulations.

Th3-7(5) 16:20-16:35 Th3-7(6) 16:35-16:50

Estimation of Bounded Unknown Input and its Multi-Objective Optimizing for Image Recovering Application to Attitude Estimation for in Compressive Sensing Accelerated Rigid Body Using GPS/INS Measurements Sheng Bi[1][2] , Ning Xi[1] and King Wai Chiu Lai[1] Department of MBE, City University of Hong Kong, Hong Kong [1] Iftikhar Ahmad, A. El Hadri, A. Benallegue School of CS, South China University of Technology [2] All are with LISV-UVSQ, Paris-France

• We propose an observer to estimate unknown • A multi-objective optimizing method is bounded time varying input of a nonlinear system. proposed for a Image Recovery in A figure here • Exponential convergence of the observer is Compressive Sensing. Accuracy and rapidity demonstrated using Lyapunov theory. are selected for optimization goal, and a • Using INS/GPS measurements the linear multi-objective fitness function is built. acceleration of a rigid body moving in 3D-space Genetic algorithm is used in the optimization is then estimated with Exponential convergence process. Finally, the result of optimization is used in a single pixel camera system. And to be used in estimation of its attitude (using Comparison of Real and For figure caption (optional), TRIAD Algorithm, QUEST or Mahony’s filter). Estimated Euler’s Angles the multi-objective optimizing method is use arial narrow 20pt font verified.

92

Session Chairs Index

&

Authors Index

IEEE ROBIO 2012 Conference Digest

Index of Session Chairs

- A - Session - J - Session Aoi, Shinya Tu2-4 Jesus, Ortiz Th1-7 Arkin, Ronald We3-1 Asmar, Daniel Th3-5 - K - Session Kagerer, Markus We2-1 Session - B - We3-1 Bianco, Corrado Guarino Lo Th3-2 Kawabata, Kuniaki Th1-4 Kheddar, Abderrahmane Th1-3 Kosuge, Kazuhiro Tu1-4 Session - C - Kurisu, Masamitsu We1-6 Capi, Genci Tu1-7 Cetinsoy, Ertugrul We2-5 Session Chao, Fei Th2-3 - L - Chen, Heping Tu3-5 Laribi, Med Amine Tu3-3 Chen, I-Ming Th2-2 Lin, Junjian Th1-6 We3-6 Lin, Zhuohua Th3-7 Chen, Weihai We3-5 Liu, Dikai Tu2-5 Chen, Xinkai We1-6 Liu, Haoting Th2-7 Coy, Johannes Alexander Tu2-5 Liu, Ying Tu2-3 Cui, Wenhuan We3-4 Liu, Yiwei Tu1-1 Liu, Yunhui We1-4 Lu, Zhenli Tu1-2 Session - D - Lueth, Tim Th1-5 Dean-Leon, Emmanuel Th3-6 - M - Session Session - E - Ma, Shugen We1-1 Eslamy, Mahdy We2-6 Martinez, Horacio We3-7 Matthias, Rene Tu3-4 Ming, Aiguo Tu2-7 - F - Session Feng, Tiantian We3-3 Session Fu, Yili Tu2-1 - N - We1-2 Nakajima, Kohei We3-2 Funato, Tetsuro Tu2-4 Nakajima, Shuro Tu2-2 Narioka, Kenichi Tu2-2 Noh, Yohan We2-7 - G - Session Gatsoulis, Yiannis Tu2-6 Session Goeller, Michael Th3-1 - O - Gouko, Manabu Tu3-6 Okada, Yuji Th2-5 Gribovskiy, Alexey Tu3-2 Ou, Yongsheng Th2-6 Gu, Jason We1-3 Guan, Jingwei Tu2-1 Session Guan, Yisheng Tu1-4 - P - Guo, Jian Th3-7 Pan, Wei Th2-4 Guo, Shuxiang We1-2 Peng, Jun Th3-4 Th2-5 Pu, Huayan Tu1-6 - H - Session Hanai, Ryo We2-7 Session Haselich, Marcel Th2-3 - Q - Hashimoto, Minoru We2-2 Qiu, Zhicheng Th1-6 We3-2 Th2-6 Hirai, Shinichi Tu1-6 Hou, Jinping Th2-7 Session Huang, Jian We2-4 - R - Rajchakit, Grienggrai Th1-1 Rekleitis, Ioannis Tu2-7 Session - I - Ren, Hongliang We3-3 Ishikawa, Masato Th3-1 Th3-4

93 IEEE ROBIO 2012 Conference Digest

Rigatos, Gerasimos We1-5 Zhou, Ran We1-3 Roppenecker, Daniel Bernd Tu3-3 Zhu, Chi We1-7 Ryuh, Young-sun We1-1 Zhu, Jin-hui Tu3-4 Zhu, Xiaorui We2-3 - S - Session Sajjad, Haider Tu1-1 Shi, Liwei Tu3-2 Smith, Chris Th2-2 Th3-3 Soltoggio, Andrea We3-6 Song, Zhibin We3-5 Sugiyama, Seiji Th3-6 Sun, Dong Th1-4 Suzuki, Toshikatsu We1-7

- T - Session Tan, Jeffrey Too Chuan Th1-7 Toming, Gert Tu3-1

- W - Session Wada, Tetsuya Th2-1 Wakita, Yujin We3-7 Wang, Chuangqi Th1-2 Wang, Han Tu2-6 Tu3-7 Wang, Junqiu Tu1-3 Wang, Kundong We2-1 Wang, Lizhen Th2-4 Wang, Michael We1-5 Wang, Nianfeng Th1-5 Wang, Zhidong Th3-3 Wu, Wenrong Th1-1

- X - Session Xu, Dan Tu3-6 Xu, De Tu3-5 Xu, Qian Th1-3 Xu, Wenfu Th1-2

- Y - Session Yang, Gi-Hun We2-2 Yang, Wei Tu2-3 Ye, Changlong Th3-5 Yokoi, Kazuhito We1-4 We2-3 Yu, Haoyong We2-6 Yu, Yong Tu1-5

- Z - Session Zeng, Dewen Th3-2 Zhan, Qiang We2-5 Zhang, Dan Tu1-5 Zhang, Guanglie Th2-1 Zhang, Jianwei Tu1-7 Zhang, Lige Tu3-1 Zhang, Xia We3-4 Zhang, Xianmin Tu1-3 Zhang, Xinzheng Tu3-7 Zhang, Yi We2-4 Zhao, Mingguo Tu1-2

94 IEEE ROBIO 2012 Conference Digest

Index of Authors

Bi, Shusheng We1-1 We1-2 Session - A - Th3-5 Abdelaziz, Benallegue We1-4 Bi, Zhiqiang Tu2-2 Abdelhafid, El Hadri We1-4 Bianco, Corrado Guarino Lo Th3-2 Abdelhakim, Chibani Th3-1 Binetti, Giulio Th1-4 Abidi, Shaukat Tu1-1 Boer, Nicholas L.de Th3-7 Abo-Ismail, Ahmed We2-7 Bolandi, Hossein We2-5 Ahmad, Iftikhar Th3-7 Bonser, Richard Th2-7 Ahmadabadi, Majid Nili Th2-2 Bottasso, Carlo Luigi Tu1-3 Aida, Kyoko We1-3 Boubakeur, Daachi Th1-5 Th3-3 Bowling, Alan Tu2-4 Akanyeti, Otar Tu3-1 Boyer, Frederic Tu1-4 Akiyama, Yasuhiro We3-5 Branson, David Tu1-2 Alici, Gursel We3-2 Brunet, Jean-Philippe Tu3-1 Alimi, Adel M. We3-6 Bu, Chunguang Th3-5 Amar, Rezoug We3-2 Burbridge, Chris Tu2-6 An, Fengwei Th1-1 Burgkart, Rainer We3-7 Andre, Vanessa We3-4 Burschka, Darius Tu1-4 Aoi, Shinya Tu2-4 Byl, Nancy Th2-1 Tu2-4 Tu2-4 Session Aonuma, Hitoshi Th1-4 - C - Aouf, Nabil We1-5 Caelli, Terry Th2-7 Aphiratsakun, Narong Th3-2 Cai, Hegao Tu2-1 Araabi, Babak Najar Th2-1 Cai, Yueri We1-1 Arai, Tamio Th2-4 We1-2 Arbulu, Mario Th3-2 Caldwell, Darwin We3-2 Arevalo, Juan Carlos Tu1-4 Th1-7 Th2-5 Caldwell, Darwin G. Tu1-2 Arkin, Ronald We3-1 Calmettes, Vincent Tu2-3 Arsicault, Marc Tu3-3 Can, Salman We1-3 Asadi, Ehsan Tu1-3 Cannan, James We2-4 Asaka, Kinji Tu3-2 Cao, Jian Tu2-3 Asl, Hamed Jabbari We2-5 Capi, Genci Tu1-7 Asmar, Daniel Tu1-1 Cerquides, Jesus We3-6 Th2-4 Cestari, Manuel Tu1-4 Th3-5 Th2-5 Augustine, Marcus Tu1-4 Cetinsoy, Ertugrul We2-5 Chaitanya, K Th1-1 Chakravarthy, Y. Kalyan Th1-1 Session - B - Chan, Chi Chong We2-2 Baba, Kouhei Th3-6 Chang, Handdeut Tu2-4 Bagherzade, Amirali Th2-1 Chao, Fei Th2-3 Bahn, Wook Tu3-7 Chao, Kenneth Tu2-1 Bai, Shaoping Tu1-4 Chathuranga, K. V. D. S. Tu3-2 Bai, Tianxiang Tu1-3 Chatty, Abdelhak We3-6 Balch, Tucker We1-1 Chau, Long-Ho Th2-1 Banchadit, Wankwan Th1-5 Chawki, Mahfoudi Th3-1 Banerji, Debajyoti Tu3-7 Chen, Bin Th1-2 Barata, Jose Th2-3 Chen, Bo We1-4 Barrios, Luenin We3-1 Chen, Chao Tu2-5 Bartolomeo, Luca Th3-7 Chen, Cheng Th3-5 Belkhiri, Ayman Tu1-4 Chen, Diansheng We1-2 Belter, Dominik Tu3-7 Chen, Gong We2-6 Beltran, Luis Miguel Th3-2 Chen, Guimin Tu1-6 Benallegue, Abdelaziz Th3-7 Chen, Heping Th2-2 Bera, Tarun Kumar Tu3-4 Chen, Hongjun Th2-2 Bertram, Torsten We2-4 Chen, Jian-An Th3-3 Bewley, Thomas R. Tu2-2 Chen, L.M. Th2-6 Bi, Sheng Th1-7 Chen, Liangliang Th3-7 Th3-7 Chen, Ling We2-7 Bi, Shuhui Th1-6 Chen, Shizhong Tu2-2

95 IEEE ROBIO 2012 Conference Digest

Chen, Weihai We2-1 Ding, Ming We3-5 We2-1 Ding, Ning Tu2-7 Chen, Wenjie We2-1 Dittrich, Frank We2-3 We2-1 Do, Ha We3-4 Chen, Xiao Th3-3 Dobritz, Martin We2-1 Chen, Xiaogang We1-6 Dong, Min Th1-7 Chen, Xijun Th2-1 Dong, Que Tu3-5 Chen, Xuechao We3-3 Dong, Wei Tu3-3 Chen, Yang Tu3-2 Dreo, Johann Tu3-1 We1-5 Du, Chong We2-1 We2-5 Du, Jianhao Tu2-7 Chen, Yen-Lun Tu3-6 Du, Juan We1-1 Chen, Yibin Tu3-4 Du, kai Th1-1 Chen, Yongquan Tu1-7 Du, Qiliang We1-7 Chen, Zhi Tu3-5 Du, Ruxu Tu3-2 Chen, Zhong We2-6 Du, Yingkui Tu1-3 Cheng, Chi-Yuan Th2-7 Du, Zhijiang Tu3-3 Cheng, Hongtai Th2-2 Duan, Feng We1-3 Chi, Wenzheng Th2-1 We1-4 Chiba, Masaya We1-7 Th1-7 Chiba, Ryosuke Th2-4 Duan, Xingguang Tu3-3 Cho, Dong-il Tu3-7 Duan, Yulin Tu2-6 Cho, Jae Uk Tu2-2 Duhaut, Dominique We3-4 Chu, Fu-Lei Tu1-1 Dung, Feng We1-7 Chung, Myung Jin Th2-3 Th2-5 Chung, Ronald Th1-7 Chung, Wing Kwong Th1-2 Session Cisneros, Rafael We1-4 - E - Clark, Adrain Tu2-6 Edamura, Kazuya We2-1 Clark, Christopher M. Tu3-1 Eggert, Simon Th2-3 Th3-4 Eiler, Kenji Lars We3-1 Claybrough, Matthieu Th3-6 Ekanayake, Samitha Th2-7 Coelen, Vincent Th3-1 Elhajj, Imad Tu1-1 Colas, Francis Tu1-3 Th2-4 Collins, Thomas We1-1 Endo, Gen We2-7 Conradt, Jorg We1-4 Escande, Coralie Th3-1 Cosentino, Sarah Tu1-1 Coy, Johannes Alexander Tu2-5 Session Cruz, Manolo STA We2-6 - F - Cui, Wenhuan We3-4 Fan, Baojie Tu1-3 Cutkosky, Mark Tu1-5 Fan, Rongfang We2-3 Fang, Chen We2-3 Farmani, Mojtaba We2-5 Session - D - Fazli, Pooyan Th3-5 Dai, Lili We1-3 Fei, Juntao Th3-6 Dambrine, Michel We2-4 Feng, Zhi-Jing Tu1-1 Dang, Quoc Viet We2-4 Feussner, Hubertus We1-3 Dasai, Ryoichi We2-7 We2-1 De la Torre, Fernando Th3-3 We2-3 de Mantaras, Ramon Lopez We3-6 Fiebrink, Rebecca Th1-5 Dean-Leon, Emmanuel Th3-6 Figueroa, Jorge Th3-4 Dean-Leon, Emmanuel Carlos We1-3 Filipescu, Adrian Tu1-7 Defay, Francois Th3-6 Fiolka, Adam We1-3 Dehghan, Seyyed Mohammad Mehdi We2-5 Forrester, Jeffery Tu3-1 Dellaert, Frank We3-1 Franceschini, Nicolas We3-3 Demiris, Yiannis We3-6 Fu, Kin Chung Denny We1-2 Deneubourg, Jean-Louis Tu3-2 Fu, Ruiqing Tu1-7 Deng, Jiaming We2-6 Fu, Yili Tu3-5 Deng, Mingcong Th1-6 Fujie, Makasatsu We2-7 Deng, Zongquan We1-6 Fujikawa, Taro Th3-6 We1-6 Fujiki, Soichiro Tu2-4 Dequidt, Antoine We2-4 Tu2-4 Desai, Ruta Tu2-1 Fujiwara, Kiyoshi We2-7 Devassy, Joan We3-1 Funato, Tetsuro Tu2-4 Deymier, Clement Tu2-3 Diab, Hilal Th1-6 Session Dillmann, Ruediger Th3-1 - G - Ding, Baijun We3-5 Gambin, Timothy Tu3-1

96 IEEE ROBIO 2012 Conference Digest

Gans, Nicholas We2-4 Hanai, Ryo We2-7 Gao, Bingtuan Th2-2 Hanazawa, Yuta We3-2 Gao, Han We2-5 Hanh, Le Duc Th1-6 Gao, Hengrui Th3-5 Hara, Susumu We3-5 Gao, Junyao We2-2 Th2-5 Gao, Peng We1-3 Harada, Kensuke We1-7 Garcia, Elena Tu1-4 Hasegawa, Yasuhisa We3-7 Th2-5 Haselich, Marcel Th2-3 Garzon, Claudia Th3-2 Hashimoto, Minoru We3-4 Gatsoulis, Yiannis Tu2-6 We3-5 Th1-1 Hasnain, Syed Khursheed We3-6 Gaussier, Philippe We3-6 Hatake, Kazuyuki Tu1-1 Ge, Jian We3-3 Hatao, Naotaka We3-1 Ge, Yunjian Tu1-5 Hattori, Takamasa Th2-5 George, Thomas Tu3-1 Hauck, Eckart Th1-6 Geyer, Hartmut Tu2-1 Hausberger, Martine We3-4 Ghilardelli, Fabio Th3-2 Hayakawa, Yoshikazu Tu1-7 Ghorbel, Yessine We2-6 Hayashibara, Yasuo Th1-3 Gillen, Sonja We2-1 He, Haiyang Tu2-5 Goeller, Michael Th3-1 He, Jiafeng Tu3-6 Gong, Zhenbang Tu1-6 He, Jiayuan We3-7 Tu3-1 He, Qing Tu2-3 Gonzalez, Daniel Th3-1 Tu3-6 Gonzalez, Jose Tu1-4 Th2-5 Gonzalez, Leonardo Th3-2 Th3-2 Gouko, Manabu Tu3-6 He, Yu Tu3-5 Gribovskiy, Alexey Tu3-2 He, Yuqing Th2-6 Griffiths, Gareth Tu1-5 Th3-5 Gu, Jason Tu1-3 Herath, Sanvidha Charaka Kumara Th3-7 Gu, Ye We3-4 Th3-7 Guan, Guan Tu2-3 Hernandez, Monica Th3-1 Tu3-6 Heyneman, Barrett Tu1-5 Th2-5 Hirai, Shinichi Tu1-6 Th3-2 Tu3-2 Guan, Jingwei Tu2-1 Hirata, Yasuhisa Tu3-7 Guan, Yisheng Tu1-5 Th2-4 Tu2-2 Th2-4 We1-4 Ho, Van Anh Tu3-2 Th3-1 Hocaoglu, Elif Th1-3 Guglielmino, Emanuele Tu1-2 Hocine, Daachi Mohamed EL Th1-5 We3-2 Hoffmann, Frank We2-4 Gumprecht, Jan D.J. We2-3 Homma, Keiko We2-7 Gumprecht, Jan David Jerome Tu2-5 Horst, Gerald Tu3-3 Guo, Bin Tu3-5 Hosoda, Koh Tu2-2 Guo, Hongwei We1-6 Tu2-4 Guo, Jian Th3-7 We2-2 Guo, Jing Th3-4 Th1-4 Guo, Shuxiang Tu3-2 Hou, Chao Th1-2 We1-1 Hou, Jinping Th2-7 We3-5 Hou, Xuyan We1-6 Th3-2 Hou, Yueen Th2-3 Th3-7 Hou, Zhi Th1-7 Hu, Biao Tu1-2 Hu, Chao We1-3 Session - H - We1-3 Hadri, Abdelhafid El Th3-7 Hu, Chongyang Tu1-4 Haghshenas-Jaryani, Mahdi Tu2-4 Tu3-4 Hagiwara, Keisuke We2-3 Hu, Huosheng Tu3-6 Haider, Sajjad Tu1-1 We2-4 Tu3-6 We2-7 We3-6 We3-7 Halloy, Jose Tu3-2 Hu, Juqi We3-1 Han, Dawei Tu3-3 Hu, Jwu-Sheng Th2-7 Han, Jianda Tu1-3 Th3-3 Th2-6 Hu, Xiaoqi We2-6 Th3-5 Hu, Ying We1-3 Han, Song Th2-1 Th1-2 Han, Teawon We3-1 Hua, Lei Tu3-5

97 IEEE ROBIO 2012 Conference Digest

Huang, Chengzu We2-2 Jiang, Zhihong Tu3-5 Huang, Han-Pang Tu2-1 Jiao, Jianbin We1-6 Huang, Haojun We3-3 Jin, Haiyang We1-3 Th2-2 Jin, Jiamei We2-6 Huang, Jian We2-4 Jin, Xuan We2-1 Huang, Jun Tu1-6 Jin, Yichen Tu2-7 Huang, Qiang Tu2-2 Johnson, Jared We3-6 Tu3-5 Jost, Celine We3-4 We3-3 Jun, Ota We3-6 Huang, Ruining Th2-2 Junginger, Steffen We1-2 Huang, Shouren Th2-6 Th1-2 Huang, Sunan We2-6 Huang, Xiaogang We1-7 Session Huang, xinhan Th1-1 - K - Huang, Yanjiang Th2-4 Kagami, Satoshi We3-1 Huang, Yuancan Th2-6 Kagerer, Markus We2-1 Huang, Zhifeng We1-3 We3-1 Th3-3 Kakogawa, Atsushi Th3-5 Huo, Xijian Tu2-1 Kallel, Ilhem We3-6 Kanai-Pak, Masako We1-3 Th3-3 Session - I - Kanbayashi, Takashi Th2-4 Iemura, Yu We3-2 Th2-4 Ietomi, Kazuhisa Tu2-2 Kang, Bo Tu2-5 Iida, Naritoshi We1-2 Kang, Jungwon Th2-3 Ikemoto, Shuhei Tu2-4 Kang, Rongjie We3-2 Imuta, Ryo Tu1-6 Kang, Xin Tu3-3 Inamura, Tetsunari Th1-7 Kantor, Geroge Th3-3 Iribe, Masatsugu We2-7 Karimi, Mahmood Tu2-2 Irlinger, Franz We3-1 Kashiwazaki, Koshi Th2-4 Ishiguro, Akio Th3-4 Katayama, Daiki Tu2-4 Ishiguro, Hiroshi We1-2 Kawabata, Kuniaki Th1-4 Ishii, Hiroyuki Tu1-1 Kawai, Yoshihiro We1-7 Tu1-1 Kiely, Janice Th1-2 We1-2 Kikuchi, Takehito Th1-7 We2-7 Th2-1 Th3-7 Kim, Chyon Hae Tu3-6 Ishikawa, Masato Th2-4 We2-2 Th3-1 Kim, Hee Joong We1-1 Th3-4 Kimura, Hiroshi We1-2 Ishikawa, Masatoshi Th2-6 Kinio, Steven Tu2-5 Ishikawa, Shun We3-5 Kinoshita, Shinichi We1-2 Isogai, Kaoru We3-5 Kinugasa, Tetsuya We2-7 Ito, Makoto We2-6 Kita, Takuto Th2-4 Ito, Soichiro Th2-5 Th3-1 Iwata, Hiroaki Th2-3 Kitajima, Yasuko We1-3 Iwata, Hiroyasu Th1-7 Th3-3 Th2-5 Kleeff, Joerg We2-1 Knoll, Alois We1-3 We2-3 Session - J - Th3-6 Jalgha, Bassam Tu1-1 Kobayashi, Keigo Th2-1 James, Russel Thomas Tu3-4 Kobayashi, Wataru Th1-7 Jang, Byung-moon Tu3-7 Kobayashi, Yuichi Tu3-6 Jensen, Brian We1-3 Kohda, Takehisa Tu2-4 We2-3 Kohrt, Christian Th1-2 Jeronimidis, George Th2-7 Koizumi, Takayuki Tu2-6 Jeschke, Sabina Th1-6 Kong, Xiangwang Tu1-7 Jia, JiaYunyi Th1-3 Kong, Xiangzhan Tu3-3 Jia, Songmin Tu2-3 Kong, Xin Tu3-6 Tu2-3 Kong, Yixiao We2-6 Tu2-7 Kosuge, Kazuhiro Tu3-7 Jiang, Guolai We2-4 Th2-4 Jiang, Li Tu2-1 Th2-4 Jiang, Min Th2-3 Kowalewski, Stefan Th1-6 Jiang, Ping Tu1-7 Kranzfelder, Michael We2-3 Jiang, Shengyuan We1-6 Krishna, P. Murali Tu1-2 Jiang, Xiao-Bao We2-2 Kruusmaa, Maarja Tu3-1

98 IEEE ROBIO 2012 Conference Digest

Kubus, Daniel Th3-2 Li, Yanfei We1-2 Kudo, Hiroki Tu3-2 Li, Yi-zhen Tu3-4 Kumar, R. Prasanth Tu1-2 Li, Zexiang Th2-2 Kumpakeaw, Saman Th2-7 Th3-6 Kurisu, Masamitsu We1-6 Li, Zheng Tu3-2 Kuwahara, Noriaki We1-3 Li, Zhijun Th2-6 Th3-3 Lian, Feng-Li Th3-3 Liang, Bin Th1-2 Liang, Jianhong Tu3-2 Session - L - We1-5 Labecki, Przemyslaw Tu3-7 We1-5 Lai, Jiun-Jau Th3-3 We2-5 Lai, King Wai Chiu Th3-7 Liang, Jinglun Tu1-5 Lam, Tinlun We3-7 Liang, Qiaokang Tu1-5 Laribi, Med Amine Tu3-3 Liang, Wenyuan Tu1-5 Laroque, Philippe We3-6 Liang, Xubin Th2-7 Lau, M.W.S. Tu3-7 Liang, Yan Tu1-7 Lee, Hyun-Jin We2-2 Liao, Baofeng Tu3-2 Lee, Jihong We1-1 Liao, Wei-Hsin We2-2 Lee, Sang-Hyo We2-2 Lin, Chuan Tu3-6 Lee, Tae-jae Tu3-7 Lin, Chyi-Yeu Th1-6 Lehr, Jane Tu3-1 Lin, Haixiong Th2-3 Lei, Xiaokang Th1-4 Lin, Junjian Th1-6 Lemasson, Alban We3-4 Lin, Peidong Th1-7 Lemme, Andre We3-6 Lin, Zhuohua Tu1-1 Leng, Bin Tu2-3 We1-2 Tu3-6 Th3-7 Th2-5 Ling, Pin-Yong Tu2-1 Th3-2 Liou, Jay Y.J. Th1-7 Levedahl, Blaine We1-4 Liu, Bo Tu3-1 Li, Baopu We1-3 Liu, Chunfang Tu1-7 We1-3 Liu, Dikai Tu1-6 Li, Bin-bin Tu3-4 We3-2 Li, Fan Th2-7 Liu, Guangliang Th1-6 Li, Geng Tu1-6 Liu, Guanyang Tu2-5 Li, Guanglin Th2-6 Liu, Haoting Th2-7 Li, Guodong Th2-6 Liu, Hong Tu1-6 Li, Heng Th1-4 Tu2-1 Th2-4 Tu2-1 Li, Hengyu We3-1 We3-4 Li, Hongyi Tu3-4 Th1-2 Th1-4 Liu, Houde Th1-2 Li, Huaizhu Th3-1 Liu, Huaxin We2-2 Li, Hui Tu3-5 Liu, Hui Th1-2 Li, Jian Th2-6 Liu, Jianwei We3-7 Li, Jing Tu3-3 Liu, Liangpeng Th1-7 Li, Luyang Tu3-7 Liu, Ming Tu1-3 Li, Maohai Tu3-7 Th3-4 Li, Min Tu3-1 Liu, Mingyong Th1-4 Li, Peng We1-6 Liu, Rongqiang We1-6 Li, Tao We3-2 Liu, Shuang Th1-3 Li, Wei We2-2 Liu, Sicen We1-3 We3-6 Liu, Tianlong We3-1 Th3-4 Liu, Tingting We1-7 Li, Weiguang Th2-3 Liu, Wei Th3-3 Li, Weihua We3-2 Liu, Weirong Th1-4 Li, Wen J. Th1-7 Th2-4 Th2-1 Liu, Xiang Tu1-1 Li, Wenbai Th1-4 Liu, Yanwei Tu1-4 Li, Wenyu We1-4 Tu3-4 Li, Xiaodong We1-5 Liu, Yi We2-2 Li, Xiaoxiao We1-3 Liu, Ying Tu2-3 Li, Xin We2-2 Liu, Yiwei Tu1-1 Li, Xinfeng Th2-7 Tu2-1 Li, Xiuzhi Tu2-3 Liu, Yongsheng We2-1 Tu2-3 Liu, Yue Th3-6 Tu2-7 Liu, Yunhui Tu3-7 Li, Y.F. Tu1-3 Liu, Zhitian Th1-7

99 IEEE ROBIO 2012 Conference Digest

Lopes, Luis Seabra Tu1-2 Meng, Max Q.-H. Th3-4 Lotfi, Benziane We1-4 Meng, Qing-Hao We2-2 Lou, Huidong Th2-3 We3-6 Lou, Yunjiang Th2-2 Th3-4 Th3-6 Menon, Carlo Tu3-3 Lu, Zhenli Tu1-2 Mertsching, Barrbel Tu2-7 Luan, Yubao We2-5 Merzouki, Rochdi Tu3-4 Lueth, Tim C. We2-1 Th3-1 We2-3 Min, Huaqing Tu3-6 We2-3 Th3-7 Lueth, Tim Christian Tu2-5 Min, Hua-qing Tu3-4 Tu3-3 Min, Huasong We3-3 We3-1 Th2-2 Luo, Jun We3-1 Minca, Eugenia Tu1-7 Luo, Ronghua Tu3-6 Ming, Aiguo We1-7 Luo, Yuan Tu3-6 Mirabdollah, Mohammad Hossein Tu2-7 Lv, Guangming Th3-5 Mitsumata, Tetsu Th1-7 Miura, Jun We1-7 Miyazaki, Yuta Th3-4 Session - M - Mizoguchi, Hiroshi We3-5 Ma, Biao Th1-6 Mizuguchi, Daichi We2-7 Ma, Hongwei We1-1 Mizuno, Haruaki Tu2-6 Ma, Qiongxiong We2-7 Mohamed, Tadjine We3-2 Ma, Shugen Tu1-6 Mok, Aloysius Th2-1 Tu3-1 Mondada, Francesco Tu3-2 We1-6 Moon, Jeon-Il We1-1 We3-2 Moradi, Hadi We2-5 Th3-5 Th2-1 Th3-5 Th2-2 Mackworth, Alan Th3-5 Mou, Wei Tu3-7 Maeda, Jukai We1-3 Mou, Xiaozheng Tu2-6 Th3-3 Murakami, Kazunori Th2-4 Mai, Xiaochun Tu2-7 Th2-4 Mair, Elmar Tu1-4 Muscio, Giuseppe Th1-6 Majima, Tatsuo Tu2-5 Mustapha, Hamerlain We3-2 Makapunyo, Teesit Th2-2 Malzahn, Jorn We2-4 Session Mamdouh, Mohamed Th1-3 - N - Mansour, Chadi Th2-4 Nagahama, Shyunsuke We2-2 Mantel, Kevin Th3-4 Nagai, Suguru Tu2-5 Markovic, Rok We1-5 Nagata, Ayanori We1-3 Martinez, Horacio We3-7 Th3-3 Masuda, Yasunobu Th1-7 Nagata, Kazuki We1-7 Matko, Drago We1-5 Nagendran, Arjun Th1-3 Matsui, Toshikazu We2-3 Nair, Suraj Th3-6 Matsuoka, Yoky Th1-5 Nakabo, Yoshihiro We2-7 Mattausch, Hans Juergen Th1-1 Nakajima, Kohei Tu1-2 Matthias, Rene Tu3-4 We3-2 Mattos, Leonardo Th1-7 Nakajima, Shuro Tu2-2 McDonald-Maier, Klaus We2-7 Th3-6 We3-7 Nakamura, Kenichi Th2-4 McGinnity, T. Martin Tu2-6 Th2-4 Th1-1 Nakamura, Mitsuhiro We1-3 McVicker, William Tu3-1 Th3-3 Mei, Tao Tu1-4 Nakamura, Taro Tu2-5 Tu3-4 Nakamura, Yutaka We1-2 Mei, Zhanyong Th1-5 Nakashima, Akira Tu1-7 Meining, Alexander Tu3-3 Nakazawa, Nobuaki We2-3 We1-3 Nandy, Sambhunath Tu3-7 Melhuish, Chris Tu1-5 Narioka, Kenichi Tu2-2 Melo, Kamilo Th3-1 We2-2 Mendonca, Ricardo Th2-3 Naso, David Th1-4 Meng, Max Q.-H. Tu2-1 Neubert, Jeremiah We3-6 Tu2-1 Ni, Huichao Th3-5 We1-3 Nihei, Yuma We3-1 We1-3 Ning, Li We3-1 We1-7 Niparnan, Nattee We3-4 Th2-1 Th2-2

100 IEEE ROBIO 2012 Conference Digest

Nishi, Yuki We2-2 Pathirana, Pubudu Nishantha Th3-7 Nishimura, Hiroyuki Th3-5 Patoglu, Volkan Th1-3 Niu, Chuanmeng We1-1 Patriciu, Alexandru Tu2-5 We1-2 Paulus, Dietrich Th2-3 Noh, Yohan Tu1-1 Peng, Jun Th1-4 We2-7 Th2-4 Noohi, Ehsan Th2-2 Peng, Shuang Tu1-2 Norzalilah, Mohamad Nor We3-2 Peng, Yi We1-6 Nour, Omar We2-7 Petersen, Klaus Tu1-1 Novitzky, Michael We1-1 Pevedic, Brigitte Le We3-4 Pfeifer, Rolf Tu1-2 We3-2 Session - O - Phoka, Thanathorn Th2-2 Oberlaender, Jan Th3-1 Phung, Anh Son We2-4 Ogata, Taiki We1-3 Pierri, Francesco Th1-6 Th3-3 Pillat, Remo Th1-3 Ogawa, Yuji Tu3-7 Pipattanasomporn, Peam Th2-2 Ogura, Kazuya We1-7 Pipe, Anthony Tu1-5 Ogure, Takuya We2-7 Th1-2 Ohba, Kohtaro We2-7 Pippin, Charles We1-1 Ohori, Suguru Th1-7 Pomerleau, Francois Tu1-3 Okabayashi, Satoshi We1-2 Porez, Mathieu Tu1-4 Okada, Yuji We1-7 Priot, Benoit Tu2-3 Th2-5 Pu, Huayan Tu1-6 Okamoto, Shogo We3-5 Th2-5 Session Okamoto, Tamao Th1-1 - Q - Okuyama, Isamu Tu1-1 Qi, Juntong Th2-6 Onda, Hiromu We1-7 Qian, Huihuan Tu1-7 Onodera, Takayuki We3-5 Tu2-7 Oriolo, Giuseppe We2-5 We3-7 Ortiz, Geman Th3-2 Qian, Jinwu We3-4 Ortiz, Jesus Th1-7 We3-5 Ortmeier, Frank Tu1-4 Th1-5 Oshida, Kotaro Th3-3 Qian, L.F. Th2-6 Osuka, Koichi Tu2-4 Qiao, Guifang We3-3 We2-7 We3-3 Th2-4 Qin, Chuanbo We1-7 Th3-1 Qiu, Chunxin We2-3 Th3-4 Qiu, Guo-Yuan Tu1-2 Ota, Jun We1-3 Qiu, Zhicheng Th1-6 Th2-4 Th2-6 Th3-3 Qiu, zurong Th1-1 Th3-4 Quan, Qiquan We1-6 Ottnad, Thomas We3-1 Ou, Yongsheng Tu2-7 Session We2-4 - R - We3-4 Rad, Ahmad B. Tu2-7 Th2-6 Rajchakit, Grienggrai Th1-1 Ouldbouamama, Belkacem Tu3-4 Rakprayoon, Panjawee Tu3-5 Ouyang, Fan Th2-2 Ramadan, Ahmed We2-7 Ouyang, Mengxing Th2-1 Th1-3 Owatchaiyapong, Peerapat Th1-5 Ran, Chongjian Th2-6 Ranasinghe, Nadeesha We3-1 Rao, K. Rajasekhara Th1-1 Session - P - Ray, Ranjit Tu3-7 Padir, Taskin Tu2-5 Raza, Saleha We3-6 Pagano, David We3-2 Raza, Syed Ali Tu3-6 Paine, Nicholas Th2-1 Rechy-Ramirez, Ericka Janet We3-7 Pan, Hongqing Th2-7 Rekleitis, Ioannis Tu2-7 Panayiotou, Christos Tu3-7 Ren, Chao We1-6 Pang, Muye We3-5 Ren, Hongliang Tu3-3 Parastegari, Sina Th2-2 Reuter, Sebastian Th1-6 Park, Jong Hyeon Tu2-2 Rezvankhah, Shayan Th2-1 Park, Sang Un Th2-3 Ribes, Arturo We3-6 Pathak, Pushparaj Mani Th3-1 Rigatos, Gerasimos We1-5 Pathirana, Pubudu Th2-7 Ristolainen, Asko Tu3-1 Pathirana, Pubudu Nishantha Th3-7 Riviere, Thomas Tu3-3

101 IEEE ROBIO 2012 Conference Digest

Rochdi, Merzouki Th3-1 Shi, Liwei We1-1 Roennau, Arne Th3-1 Shi, Qing We1-2 Rojas, Juan We1-7 Shi, Yun Tu2-6 Roppenecker, Daniel B. We2-3 Shi, Zhidong We2-3 Roppenecker, Daniel Bernd Tu2-5 Shibasaki, Ryosuke Tu2-6 Tu3-3 Shimojo, Kazuma Th1-3 Rosendo, Andre Tu2-2 Shimojo, Makoto We1-7 We2-2 Shirafuji, Shouhei Tu2-4 Rossiter, Jonathan Tu1-5 Shoji, Satoru Tu1-1 Roubieu, Frederic L. We3-3 We2-7 Ruchanurucks, Miti Tu3-5 Shome, Sankar Nath Tu3-7 Ruffier, Franck We3-3 Shota, Simazu Th2-5 Ryuh, Young-sun We1-1 Shuang, Feng Th2-7 Ryuh, Young-Sun We2-2 Siegwart, Roland Tu1-3 Silva, Filipe Tu1-2 Simazu, Shota We1-7 Session - S - Sinyukov, Dmitry Tu2-5 Saber, Mefoued Th1-5 Siradjuddin, Indrazno Th1-1 Saito, Hajime We2-7 Skrzypczynski, Piotr Tu3-7 Sakuyama, Takuya Th3-4 Soltoggio, Andrea We3-6 Salumae, Taavi Tu3-1 Song, Aiguo We3-3 Samantaray, Arun Kumar Tu3-4 Song, Guangming We3-3 Samejima, Ippei We3-1 We3-3 Samer, Mohammed Th1-5 Song, Ho-Jeong Tu3-7 Sams, Benjamin A. Tu2-2 Song, Quanjun Th2-7 Saneyoshi, Keiji We2-6 Song, Shuang We1-3 Th2-3 Song, Zhan Tu2-3 Th3-3 Th1-7 Sang, Lingfeng Tu1-7 Song, Zhibin We3-5 Sang, Weiwei We2-3 Th3-2 Sankai, Yoshiyuki Tu2-4 Soni, Bhuman Tu2-6 Santana, Pedro Th2-3 Sousselier, Thomas Tu3-1 Sanz-Merodio, Daniel Tu1-4 Sowmya, Arcot Tu2-6 Th2-5 Sproewitz, Alexander Tu2-2 Satake, Junji We1-7 Srinath, A Th1-1 Schiedermeier, Gudrun Th1-2 Srivastava, Suresh Tu1-2 Schneider, Armin We1-3 Stamp, Richard Th1-2 We2-3 Staub, Christoph We1-3 Schwaiger, Johannes We2-1 Stavrou, Demetris Tu3-7 Seet, Gerald Tu3-7 Steil, Jochen J. We3-6 Segura-Meraz, Noel We3-7 Stelzer, Annett Tu1-4 Sekine, Masashi We3-5 Stoll, Norbert We1-2 Senda, Kei Tu2-4 Th1-2 Tu2-4 Strauss, Gero We2-3 Senoo, Taku Th2-6 Su, Chunyi Th2-6 Sentis, Luis Th2-1 Su, Manjia Tu1-5 Serres, Julien We3-3 We1-4 Sessa, Salvatore Tu1-1 Su, Xuandong We2-2 Th3-7 Subramanian, Saravanakumar Tu3-1 Shammas, Elie Tu1-1 Suda, Hiroyuki We3-2 Th2-4 Sudsang, Attawith We3-4 Th3-5 Th2-2 Shao, Xiaowei Tu2-6 Sueoka, Yuichiro Th2-4 Sharif, Muhammad Usman Tu3-6 Th3-4 She, Yu Th1-2 Suga, Yuki Th1-7 Shen, Huiping We2-6 Sugahara, Yusuke Th2-4 Shen, Jinglin We2-4 Th2-4 Shen, Linyong We3-4 Sugano, Shigeki We2-2 We3-5 Th1-7 Th1-5 Th2-5 Shen, Qi Tu3-2 Sugimori, Kento We3-5 Shen, Wei-Min We3-1 Sugimoto, Yasuhiro Tu2-4 Shen, Wenhao We3-1 Th2-4 Shen, Xiaowen Tu3-6 Th3-4 Sheng, Weihua Tu2-7 Sugita, Hikaru We1-2 We3-4 Sugita, Yoshihisa Tu1-1 Shi, H.J. Th2-6 Sugiyama, Masao Th2-1 Shi, Liwei Tu3-2 Sugiyama, Seiji Th3-6

102 IEEE ROBIO 2012 Conference Digest

Sukmanee, Wantana Tu3-5 Tokumoto, Mitsuhiro Tu1-1 Sukwan, Tassanai Th1-5 We2-7 Sumioka, Hidenobu Tu1-2 Tomazic, Tine We1-5 Sun, Dong Th1-4 Toming, Gert Tu3-1 Sun, Hongtao We3-3 Tomita, Nozomi Tu2-4 We3-3 Tomiyama, Ken Th1-3 Sun, Jianquan Tu1-7 Tomizuka, Masayoshi Th2-1 Sun, Kuan-Chun Th2-7 Tomokuni, Nobuyasu We2-4 Sun, Shaoming Tu1-4 Tomori, Hiroki Tu2-5 Tu3-4 Torres-Jara, Eduardo Tu2-5 Sun, Xiantao We2-1 Traeger, Mattias We2-1 Sun, Xin Th2-7 Traeger, Mattias F. We2-3 Sun, Yi Tu1-6 Tseng, Chin-Yuan Th3-3 Sun, Yuandong Tu2-7 Tsuchiya, Kazuo Tu2-4 Sun, Yue We2-1 Tu2-4 Sun, Zhe We1-3 Tu2-4 Suppa, Michael Tu1-4 Tsujiuchi, Nobutaka Tu2-6 Suratno, Hendra Tu3-7 Tsukada, Natsumi We1-7 Suthakorn, Jackrit Th1-5 Turchiano, Biagio Th1-4 Th3-2 Suzuki, Koki Th2-4 Session Th2-4 - U - Suzuki, Tatsuya We1-4 Udsatid, Pakorn We3-4 Suzuki, Toshikatsu We1-7 Ueno, Shouhei We2-1 Th2-5 Ueyama, Tsuyoshi Th2-4 Uhl, Klaus Th3-1 Ulbrich, Heinz Tu3-3 Session - T - We3-7 Taha, Chettibi Th3-1 Takahashi, Junji We3-7 Session Takanishi, Atsuo Tu1-1 - V - Tu1-1 Vermeiren, Laurent We2-4 We1-2 Viollet, Stephane We3-3 We2-7 Visentin, Francesco Tu3-1 We2-7 Vivet, Damien Tu2-3 Th3-7 Voda, Alina Tu1-7 Takayama, Toshiyuki We2-7 von Deimling, Constantin We3-7 Takemura, Hiroshi We3-1 von Eisenhart-Rothe, Rudiger We3-7 We3-5 Takemura, Kenjiro We2-1 Session Takubo, Toshio We2-7 - W - Tan, Jeffery Too Chuan We1-3 Wada, Kazuki We2-4 We1-4 Wada, Kazunori We2-3 Tan, Jeffrey Too Chuan Th1-7 Wada, Tetsuya Tu3-7 Tan, Min Tu3-2 Wakita, Yujin We3-7 Tang, Cheng We2-4 Waldron, Kenneth We3-2 Tang, Dewei We1-6 Wang, Bin Tu1-1 Tang, Shichuan Th3-5 Wang, Chaolei We1-5 Tang, Yandong Tu1-3 We1-5 Tang, ZhenJin Th2-5 We2-5 Tanino, Tetsuya We2-4 Wang, Cheng We2-2 Tao, Erpan We3-1 Wang, Chuangqi Th1-2 Tao, Yanbo We3-7 Wang, Chunbao Tu1-1 Tarvainen, Tapio We3-5 We2-7 Tarvainen, Tapio Veli Juhani Tu1-4 Wang, Cong We1-5 Tavakolan, Mojgan Tu3-3 Wang, Danwei Tu3-7 Tazaki, Yuichi We1-4 Wang, Haibin Th2-5 Techakittiroj, Kittiphan Th3-2 Wang, Han Tu2-6 Temram, Auchara Th1-5 Tu3-7 Terunaga, Chihara We2-7 Wang, Hongbo Tu1-7 Terunage, Chihara Tu1-1 Wang, Jing Th1-4 Than, Trung Duc We3-2 Th2-4 Thondiyath, Asokan Tu3-1 Wang, Junqing We2-5 Thurow, Kerstin We1-2 Wang, Junqiu Tu1-3 Th1-2 Wang, Kai Tu3-7 Tian, Lianfang We1-7 Wang, Kundong We2-1 Tian, Xin Tu1-7 Wang, Lei Tu3-1 Toh, Siew-Lok We2-6 Wang, Liang We2-6

103 IEEE ROBIO 2012 Conference Digest

Wang, Lujia Th3-4 Wu, Yunjie We2-1 Wang, Nianfeng We1-6 Wu, Yu-Xiu Th3-4 Th1-5 Wu, Zhengxing Tu3-2 Wang, Qinruo Tu1-2 Wang, Sen We2-7 Session Wang, Tianmiao Tu3-2 - X - We1-2 Xi, Ning Tu3-4 We1-5 Th1-3 We1-5 Th3-7 We2-5 Xi, Xi We2-5 Wang, Tianran Th2-6 Xiang, Jingyu We1-4 Wang, Wei We1-2 Xiang, Kui Th3-2 Th1-7 Xiao, Jianming Th1-4 Th2-7 Th2-4 Wang, Weiguo We3-3 Xiao, Jizhong We2-5 We3-3 Xiao, Nan Th3-7 Wang, Wenmin We3-4 Xiao, Zhiguang Tu1-5 Wang, Yan Tu2-5 Xie, Dawei We1-2 Wang, Yang We2-2 Xie, Qunqun We2-4 Wang, Yaonan Tu1-5 Xie, Shaorong We3-1 Wang, Yaoxiong Th2-7 Xie, Zongwu Tu1-1 Wang, Yiyu Th3-2 Tu2-1 Wang, Yongfeng Th1-5 Xin, Jiayi Th1-7 Wang, Yonggui Tu3-3 Xing, Chuhao Tu1-4 Wang, Yubing Th2-7 Xu, Binbin We1-3 Wang, Yuechao Th1-4 Xu, Dan Tu3-6 Wang, Yunliang Th3-7 Xu, Guoqing Tu2-3 Wang, Zheng Th3-6 Xu, Hongwei Tu2-3 Wang, Zhidong Tu3-7 Tu3-6 Wang, Zhimin Tu1-2 Th2-5 Ward, Peter Tu1-6 Th3-2 Weede, Oliver We2-3 Xu, Jiali Th1-5 Wei, Bo Tu3-5 Xu, Jinkai We2-1 Wei, Hongxing We3-3 Xu, Min Tu3-1 Th2-2 Xu, Qian We3-3 Wei, Jian-Long We3-6 Xu, Wenfu Th1-2 Wei, Shuai We1-6 Xu, Xinli Tu3-6 Weikersdorfer, David We1-4 Xu, Y.D. Th2-6 Wen, Mengfei Th1-4 Xu, Yangsheng Tu1-7 Weng, Liyuan Tu3-1 Tu2-7 West, Michael We1-1 We2-4 Wicaksono, Indra Bagus Th1-1 We3-7 Wilhelm, Dirk We2-3 Th1-2 Williams, Gareth Th3-7 Xu, Zhe We2-2 Williams, Mary-Anne Tu1-1 Th1-5 We3-6 Xue, Jianru Th1-4 Winstone, Benjamin Tu1-5 Xun, Binbin We1-4 Wood, Zoe J. Tu3-1 Woodard, Ky Tu2-2 Session Worn, Heinz Tu3-4 - Y - We2-3 Yacine, Amirat Th1-5 Wu, Jia Yun Th3-3 Yadav, Jitendra Th3-2 Wu, Junjun Tu1-5 Yagi, Yasushi Tu1-3 We1-4 Yamada, Shinya Tu2-6 Wu, Lijun We1-6 Yamada, Yoji We3-5 Wu, Qianying We2-2 Th1-1 Wu, Shih-Hung Tu1-2 Th2-5 Wu, Wenqiang Tu1-5 Yamakawa, Yuji Th2-6 Th3-1 Yamakita, Masaki We3-2 Wu, Wenrong Th1-1 Yamamoto, Tomoyuki We1-2 Wu, Xiaoling Tu2-3 Yamanobe, Natsuki We1-7 Wu, Xinyu Tu1-7 We3-7 Tu2-3 Yan, Ci We3-6 Tu3-6 Yan, Fei We2-5 Th1-7 Yan, Jiu-Lou Tu2-1 Wu, Xuan Tu1-4 Yan, Xiao Th1-4 Tu3-4 Yan, Yuling We1-7 Wu, Yongqiang Th1-5 Th2-5

104 IEEE ROBIO 2012 Conference Digest

Yang, Fangping Th1-4 Zeng, Dewen Tu3-6 Yang, Gi-Hun We2-2 Th2-5 Yang, Jie Tu3-1 Th3-2 Yang, Jun-Yu Th3-3 Zeng, Ming We2-2 Yang, Ningjia We1-3 We3-6 Yang, Panpan Th1-4 Th3-4 Yang, Tingli We2-6 Zhai, Jingmei Tu2-5 Yang, Wei Tu2-3 Zhan, Qiang We2-5 Yang, Wei-Xing Th3-4 Zhang, Dan Tu1-5 Yang, Wenlong Tu3-3 Zhang, Dianfan Tu1-7 Yang, Xingbang Tu3-2 Zhang, Dingguo We3-7 Yang, Yong Tu1-7 Zhang, Fuhai Tu3-5 Yao, Guocai Tu3-2 Zhang, Guanglie Th1-7 Yasin, Awais Tu2-2 Th2-1 We3-3 Zhang, Hong Tu2-2 Yau, Waipan Tu3-3 We1-4 Ye, Changlong We1-6 Th3-1 Th3-5 Zhang, Huifeng We3-4 Ye, Guoqiang Th2-3 Zhang, Jianbin We2-1 Ye, Qixiang We1-6 Zhang, Jianfen Tu2-7 Yeon, Je Sung Tu2-2 Zhang, Jianhui Tu1-6 Yi, Chang'an Tu3-6 We2-6 Yimyam, Panitnat Tu2-6 Zhang, Jianwei We1-3 Yin, Hongbo We2-6 Th1-2 Yin, Junmao We1-2 Zhang, Jing We1-6 Yin, Xiaolin Tu2-7 Zhang, Jun We3-3 Yin, Yuehong We2-6 We3-3 Ying, Can Th2-2 Zhang, Liang Th2-2 Yokoi, Kazuhito We1-4 Zhang, Lige We1-1 Yokota, Shinichi We2-1 We1-2 Yonezawa, Naoaki Th2-4 Zhang, Peng We1-3 Yoon, Woo-Keun We3-7 Th1-2 Yoshida, Eichi We1-7 Zhang, Qi Tu1-1 Yoshida, Eiichi We1-4 Tu2-1 Yoshikawa, Tsuneo Th3-6 Zhang, Qin Tu3-5 Yoshikawa, Yuichiro We1-7 Zhang, Qinghua Tu1-5 Th2-5 Zhang, Rong Th3-5 Yoshioka, Masataka We1-7 Zhang, Shenglei Th3-6 Th2-5 Zhang, Shiwu Tu3-1 Yoshitsugu, Kamiya Tu3-5 Zhang, Songyuan We3-5 Yu, dahai Th1-1 Zhang, Tie We2-7 Yu, Guoping We3-3 Th1-6 Yu, Haoyong We1-7 Th2-2 We2-6 Zhang, Weimin Tu2-2 Th2-5 Zhang, Weiwei We1-6 Yu, Huadong We2-1 Zhang, Wenlong Th2-1 Yu, Junzhi Tu3-2 Zhang, Xia We3-4 Yu, Tongzhu We2-6 We3-5 Yu, Wenwei Tu1-4 Zhang, Xiangtong Th1-6 We3-5 Zhang, Xianmin Tu1-5 Yu, Xiaoxiang Tu1-2 We1-6 Yu, Yong Tu1-5 We2-6 Yu, Yuanlong Tu1-3 Th1-5 Yu, Zhangguo Tu2-2 Zhang, Xiaohua Th2-2 We3-3 Zhang, Xinzheng Tu2-7 Yu, Zhanjiang We2-1 Zhang, Xiong Tu1-7 Yuan, Jing We1-4 Zhang, Xuchong We2-6 Yuan, Mei We2-1 Zhang, Yanan We3-4 Yue, Chunfeng We1-1 We3-5 Yusuke, Matsuoka Tu1-1 Th1-5 We2-7 Zhang, Yi Tu3-6 Zhang, Yicheng We1-5 We1-5 Session - Z - We2-5 Zecca, Massimiliano Tu1-1 Zhang, Yongjie Tu1-4 Th3-7 Tu3-4 Zeghloul, Said Tu3-3 Zhang, Yun Tu1-2 Zeng, Dewen Tu2-3 Zhang, Yunqiang Th2-2

105 IEEE ROBIO 2012 Conference Digest

Zhang, Zeming Th1-7 Zhao, Chen We2-3 Zhao, Guoru Th1-5 Zhao, Hanwang We2-3 Zhao, Huijing Tu2-6 Zhao, Jingdong Tu1-6 Zhao, Jingshan Tu1-1 Zhao, Kai We1-2 Zhao, Mingguo Tu1-2 Zhao, Qian Tu1-2 Zhao, Xue Tu2-3 Tu2-3 Zhao, Yanguo Th1-7 Zhao, Zhengyang We2-2 Zhao, Zhijun Tu1-6 Zhao, Zhili Th2-6 Zheng, Minhua Tu2-1 Zheng, Nanning Th3-3 Zheng, Weimin Tu2-3 Tu3-6 Th3-2 Zheng, Weiming Th2-5 Zhong, Xu We3-3 Zhong, Zhipeng Tu3-6 Zhou, Changle Th2-3 Zhou, Feng Th1-4 Th2-4 Zhou, Hao We3-2 Zhou, Huiyu Tu3-6 Zhou, Jian Th3-6 Zhou, Ran We1-3 Zhou, Rui We2-1 Zhou, Shengli Th1-7 Zhou, Xiaolong Tu1-3 Zhou, Yi We1-5 We2-5 Zhou, Yimin Tu2-3 Zhou, Yu We3-3 Zhou, Yumei We3-1 Zhu, Chi We1-7 We2-6 Th2-5 Zhu, Haifei Tu1-5 Tu2-2 Th3-1 Zhu, Jin-hui Tu3-4 Zhu, Liangchang Th1-7 Zhu, Xiangyang We3-7 Zhu, Xiaorui We2-3 Zhu, Xiuming Th2-1 Zhu, Zhiqiang Th2-6 Zhuang, Xuekun Th2-7 Zhuang, Yan We2-5 Zou, Yanbiao We2-7

106

Top View