industrial activities

A Suite of Ontologies for and Automation By Sandro Rama Fiorini, Julita Bermejo-Alonso, Paulo Gonçalves, Edison Pignaton de Freitas, Alberto Olivares Alarcos, Joanna Isabelle Olszewska, Edson Prestes, Craig Schlenoff, S. Veera Ragavan, Signe Redfield, Bruce Spencer, and Howard Li

ne of the basic requirements common set of terms and definitions, CORA aims to describe what a for any type of robot com­ allowing for unambiguous ­knowledge is and how it relates to other concepts. It munication, whether with transfer among any group of human, defines four big broad entities: robot Oother or humans, is robots, and other artificial systems. The part, robot, complex robot, and robotic the need for a common vocabulary group grew quickly; it is now com­ system. The term robot may have as along with clear and concise definitions. posed of 175 members representing 23 many definitions as there are authors With the growing complexity of tasks countries from a mixture of approxi­ writing about the subject. The inherent that robots are expected to perform mately 50% educational institutions, ambiguity in this term might be an issue as well as the need for multirobot 25% private companies, and 25% gov­ when one needs to specify an ontology and human–robot collaboration, the ernment entities. for a broad community like ours. We need for a standard and well-defined One of the main outputs of this acknowledge this ambiguity as an knowledge­­ representation is becom­ group is the IEEE Standard Ontologies intrinsic feature of the domain and, ing more evident. for Robotics and Automation, which is therefore, we decided to elaborate a def­ The IEEE Standard Association’s composed of a core ontology called inition based purely on necessary con­ Robotics and Automation Society Core Ontologies for Robotics and Auto­ ditions without specifying sufficient (RAS) recognized this need, and, mation (CORA) [4]. This standard conditions. Thus, CORA covers all enti­ in 2011, created specifies the main, most general con­ ties that the community­ considers as a the Ontologies cepts, relations, and axioms of robotics robot, at the cost of classifying some One of the basic for Robotics and automation (R&A), and serves as a entities as robots that may be counterin­ requirements for and Automation reference for knowledge representation tuitive to some roboticists. However, the (ORA) working and reasoning in robots as well as a for­ concepts in our ontology could be any type of robot group. The goal mal reference vocabulary for communi­ ­specialized according to the needs of communication, of the group cating knowledge about R&A between specific subdomains or applications is to develop a robots and humans. This standard was of R&A. whether with other standard to pro­ awarded the prestigious Emerging CORA was developed to be a high- robots or humans, vide an overall Technology Award by the IEEE Stan­ level standard on which domain-­specific ontology and dards Association in December 2015. efforts could build. The approach was to is the need for a associated me­ The standard was also mentioned in the define concepts in CORA that were common vocabulary thodology­­ for National Artificial Intelligence (AI) generic to all robot domains, and then knowl­­edge rep­ Research and Development Strategic these domains could specialize these along with clear and resentation and Plan, released by President Obama in concepts to address their specific infor­ concise definitions. reasoning in ro­ October 2016 [5]. This strategic plan mation requirements. When the work­ botics and auto­ focuses on the role of AI, machine ing group that developed CORA was mation to­­gether learning, automation, and robotics in created, based on the interests of the with the representation of concepts in addressing complex national problems. working group members, we expected an initial set of application domains. With the release of this standard, the subgroups to emerge that would spe­ The standard provides a unified way of CORA working group completed its cialize the concepts represented in representing knowledge and provides a task and was required to disband. How­ CORA. Currently, we expect that ever, many of the working group mem­ groups will be organized in two layers

Digital Object Identifier 10.1109/MRA.2016.2645444 bers remain involved in this work by (Figure 1). We expect that middle- Date of publication: 29 March 2017 focusing on subgroups. layer groups will develop ontologies

8 • IEEE ROBOTICS & AUTOMATION MAGAZINE • MARCH 2017 ­regarding transversal notions in robot­ ics. Lower-layer groups will concern Upper Ontology (CORA) ­specific domains in robotics and will use concepts from CORA and middle- layer groups. Robot Task Subgroup The Robot Task Representation Subgroup The Robot Task Representation sub­ group will develop a broad standard Autonomous Robot that provides a comprehensive ontology Subgroup Subgroup for robot task structure and reasoning. In this context, task refers to the con­ Figure 1. The structure of IEEE ORA efforts. The subgroups in gray are now active; the ones crete decomposition from goal to sub­ in white are planned. goals that enables the human or robot to accomplish the outcome at a specific application domains (e.g., in manufac­ to control­ multiple elements­ of the instance in time. To accomplish this, turing) where robot task representa­ manufacturing process. there is a need for a standard providing tion could be useful. The ontology Our work plan for developing the an explicit knowledge representation for provides a unified way to represent standard has two aspects. The first is to robot tasks. knowledge about robot tasks by shar­ develop the task ontology, extending This work will be a supplement to ing common representations and pre­ CORA and capturing vocabularies for the existing standard CORA ontology. serving semantic meaning. It can be robot task representation by require­ This supplement will include the pre­ utilized in manufacturing control ments analysis and surveying the litera­ sentation of concepts in an initial set of applications, where the system needs ture. The final decision-making on

®

Colour of the Logo = Pantone 347c

Black outer layer Yellow center layer shows some Red inner layer shows excessive shows no wear wear; prepare for maintenance wear; maintenance needed

Built-in wear indication reduces unplanned downtime

PMA XR90 Multilayer Conduit provides superior cable protection for robotic and automation applications. Not only does it feature an extremely long service life of up to 10,000,000 movement cycles, but it also “shows” users how much wear the conduit has already experienced. • Visual inspection capability; enables efficient maintenance planning • Reduces costs by reducing unplanned downtime • A complete solution package with solid, one-piece support brackets and half shells For more information, contact your T&B/ABB rep.

www.tnb.com/pma

© 2016 Thomas & Betts Corporation. All rights reserved.

MARCH 2017 • IEEE ROBOTICS & AUTOMATION MAGAZINE • 9 Entity

Physical Abstract

Object Process Proposition

Content Bearing Object Agent Artifact

Device Association RA

Robot Architecture Document Robot Robot Architecture Element RA Conforms Abstract Part

Abstract Part Layer RA Module RA Structure Agent Conforms Contains Information Robot Structure Robot Behavior Robot Function

Figure 2. ROA basic concepts and relationships [3].

vocabularies will be achieved through ables the community to build up a aims to create a standard ontology consensus between different group mem­ shared catalog of tasks and capabilities that specifies the domain knowledge bers. The second aspect is to develop along with their relationships (based needed to build autonomous robots, a task repository, which will provide on elements within the task ontology). operating in the air as well as in ground a set of instances that could be used The purpose of the overall standard and underwater environments. Bene­ for robotic im­ is to ensure common representations fits include ­having a common knowl­ plementation and frameworks when tasks are de­ edge base for the development and The standard and validation of scribed, so the knowledge represented integration of systems from different the task ontology. in the task ontology defines the struc­ manufacturers, which will enable in­ provides a unified The task on­ ture and content of the tasks in the teroperability and catapult design to a way of representing tology formally task repository.­ new level. defines what a Although the task ontology will The AuR-SG intends to extend knowledge and task is, and spe­c­ be the official standard when com­ CORA and define domain-specific provides a common ifies the proper­ ­ pleted, the task repository is neces­ concepts and axioms. The AuR-SG ties of tasks, the sary to help ­validate the standard and will analyze the various autonomous set of terms and properties of to provide an avenue that makes the robots (e.g., flying, ground, underwa­ definitions, allowing the hierarchy in standard more useful and practicable ter) to identify the components neces­ which tasks are in the industry. sary to endow robots with autonomy, for unambiguous placed, and the including the needed hardware and knowledge transfer ways in which The Autonomous software. To develop the standard the performance Robot Subgroup ontology for autonomous systems, among any group of the capabili­ In 2016, the IEEE RAS Standing Com­ the AuR-SG has adopted the follow­ of human, robots, ties required to mittee approved the Autonomous ing approach: accomplish the Robot (AuR) subgroup as a study group 1) Develop standard vocabularies for and other artificial tasks are mea­ (SG) to determine whether sufficient architectural concepts in IEEE 1471/ systems. sured. The task interest and resources exist to develop IEC 42010 [with a focus on Robot repository en­ an IEEE draft standard. The AuR-SG Parts (RParts)].

10 • IEEE ROBOTICS & AUTOMATION MAGAZINE • MARCH 2017 2) Develop a functional and formal [2] S. R. Fiorini, J. L. Carbonera, P. Goncalves, V. A. Vosughi, and H. Li, “Requirements for building ontology for R&A. A. Jorge, V. F. Rey, T. Haidegger, M. Abel, S. A. an ontology for autonomous robots,” Ind. Robot 3) Check/validate relationship concepts, Redfield, S. Balakirsky, V. Ragavan, H. Li, C. Int. J., vol. 43, no. 5, pp. 469–480, Aug. 2016. using function as a basis for relation Schlenoff, and E. Prestes, “Extensions to the core [4] Core ontologies for robotics and automation checking. ontology for robotics and automation,” Robot. (CORA) standard. [Online]. Available: http://­ 4) Use the developed vocabularies and Comput. Integr. Manuf., vol. 33, pp. 3–11, standards.ieee.org/findstds/standard/1872­ -2015 ontology for the conceptual design June 2015. .html of sample robot applications by [3] B. Bayat, J. Bermejo-Alonso, J. L. Carbonera, [5] The national AI research and development extending concepts of IEEE 1471/ T. Facchinetti, S. Fiorini, P. Goncalves, V. Jorge, strategic plan 2016. [Online]. Available: https:// IEC 42010 [1] and CORA (focus on M. Habib, A. Khamis, K. Melo, B. Nguyen, J. I. www.whitehouse.gov/sites/default/files/whitehouse­ _ RParts) [2]. Olszewska, L. Paull, E. Prestes, S. V. Ragavan, S. files/microsites/ostp/NSTC/national_ai_rd_strategic_ As part of the architectural con­ Saedi, R. Sanz, M. Seta, B. Spencer, M. Trentini, plan.pdf cepts development, the AuR-SG has proposed the Robot Architecture (ROA) ontology (Figure 2) to define the architectural concepts and rela­ tions. Envisaged as a conceptual framework, ROA will facilitate infor­ mation exchange about robot archi­ tectures between human and robots. Meet the State of the Art Users of ROA will instantiate ROA concepts to represent information in Intelligent Research about specific as well as generic architectures. For example, architec­ Platforms. tural diagrams in robot literature can be thought of as instances of ROA. The conceptualization described Pioneer LX below is akin to the metamodels of concept representation languages, The Pioneer LX is everything you need to get your research rolling. such as Unified Modeling Language (UML). Once the concepts, architec­ For over 15 years, Omron Adept MobileRobots has served the academic community with intelligent, ture, and core components­ for auton­ rugged, research robots. We are proud to offer our omous systems are described, case newest robot, the Pioneer LX. Based on the Adept studies will be developed. The AuR- Lynx platform, the Pioneer LX combines the reliable The Pioneer LX includes: operation of an industrial platform with the develop- SG is currently working on defining Integrated On-board Computer ment tools and support of a Pioneer research robot. key concepts to address the function­ • Intel D252 64-bit Dual Core 1.8 GHz Atom As with other MobileRobots research platforms, al features and ­capabilities of auto­ • Wireless Ethernet Communications developers can easily access and program the robot‘s • 2GB DDR3-1066 RAM nomous robots, such as behavior, on-board computer. Software libraries are provided • Available with Windows or Linux OS function, goals, and tasks, and the to speed the development of custom applications, and • 16 In /16 Out Digital I /O ­relations among them. accessible I /O and power supplies make hardware • 4 In /4 Out Analog I /O integration easy. With included guidance and Dedicated Robot Control System autonomous navigation software, the Pioneer LX • Autonomous Navigation and Acknowledgments can be tasked with up to 13 hours of autonomous Mapping Software We wish to thank the following Auton­ navigation, carrying up to 60 Kg of sensors, effectors, • SICK S300 Laser Scanner omous Robot group members who also or custom accessories. • Front Bumper Sensors • Front and Rear Sonar Sensors joined the group’s discussion: Marcos Color LED Status Indicator Rings Pioneer LX MobileRobot • Ennes Barreto, Joel Luís Carbonera, with optional camera • Accessory Mounting Deck (removable) Abdelghani Chibani, Tullio Facchinetti, accessory. • Removable Exterior Body Panels • Docking station for Autonomous or

Antonella Ferrara, Maki Habib, Vitor Manual Charging A.M. Jorge, Alaa Khamis, Bao Nguyen, View Pioneer LX Don’t forget to follow us online and Ricardo Sanz. videos at youtube.com / user/mobilerobots References [1] M. Maier, D. Emery, and R. Hilliard, “Soft­ ware architecture: ­Introducing IEEE Standard 603.881.7960 mobilerobots.com [email protected] 1471,” Computer, vol. 34, no. 4, pp. 107–109, © 2016 Omron Adept MobileRobots, LLC., 10 Columbia Drive, Amherst, NH 03031 Apr. 2001.

MARCH 2017 • IEEE ROBOTICS & AUTOMATION MAGAZINE • 11