Test-beds Rovers for Planetary Surface Exploration

Takashi Kubota*, Masatsugu Otsuki*, Takanobu Shimada*, Yoji Kuroda**, Yasuharu Kunii***

*Institute of Space and Astronautical Science, JAXA, Japan **Mechanical Engineering Department, Meiji University, Japan ***Electrical and Electronics Engineering Department, Chuo University, Tokyo, Japan e-mail: kubota@isas.jaxa.jp

Abstract Kaguya (SELenoligical Engineering Explorer), a lunar global remote sensing mission. One of main missions for Lunar or planetary exploration missions including lunar robotics exploration in post SELENE missions is to landers and rovers are earnestly under studying in Japan. demonstrate the technologies for lunar or planetary One of main missions for lunar robotics exploration is to surface exploration and human activities on the moon in demonstrate the technologies for lunar or planetary the near future. They will cover pin-point landing surface exploration. They will cover landing technology technology, reliable landing scheme with obstacle and surface exploration rover technology. Lunar geologic avoidance, safe landing mechanism on rough terrain, survey will be also performed for utilization and exploration rover, tele-science and tele-operation scientific investigation of the moon. The working group technology, automated construction etc. The following has been conducting the feasibility study of advanced top science will also be conducted in the robotics mission. technologies for lunar or planetary robotics exploration. The working group was also established in 2008 in Japan Unmanned mobile robots are expected for surface to study Japanese Mars exploration. In the preliminary exploration of the moon or planets, because mobile study, two orbiters and some landers cooperatively robots can travel safely over a long distance. To establish explore Mars. Some explorers, such as surface the rover technology, test-beds rovers have been exploration rovers, wide area exploration by airplanes, developed. This paper presents system overviews of subsurface exploration by mole type robots are under developed test-bed roves, guidance and navigation study. schemes, smart manipulators and some experimental The working group has been conducting the results . feasibility study of advanced technologies for lunar or planetary robotics exploration. Unmanned mobile robots 1 Introduction are expected for the detailed surface exploration of the moon or Mars, because rovers can travel safely over a In Japan, lunar or planetary exploration has received long distance and observe what to see by some scientific a lot of attention. JAXA completed Kaguya and instruments. Therefore the rover R&D group developed Hayabusa missions. Kaguya has performed the highly innovative test-bed rovers with a new mobility system, accurate and high-resolution observation of elementary lightweight manipulators, and advanced guidance and and miner compositions of surface, topography, navigation functions. The developed test-beds have a subsurface structure, magnetic anomaly and gravitational new suspension system, which consists of a four-wheel field. The asteroid sample return mission, Hayabusa drive suspension system and two actives wheels. The observed the attractive asteroid Itokawa in 2005 and proposed system is designed to distribute the load of returned back to the Erath in 2010. JAXA is promoting a weight equally to all six wheels whenever the rover Venus orbiter mission, Akatsuki, which was launched in climbs up or down, and then provides high degree of June 2010. Akatsuki will elucidate the mechanism of the mobility for the rover. Smart manipulators with a new super-rotation, the circulation of the Venusian end-effector are also developed to perform the in-situ atmosphere at speed 60 times higher than the rotation analysis or direct observation on the surface. The speed of the planet, and will investigate and identify the developed end-effector has two kinds of functions, movement of the atmosphere by infrared and other gripping and scooping. The experimental results for observations and will investigate the distribution of sample collection show the effectiveness of the lightning discharges. developed end-effector. The test-bed rovers install a Recently, new lunar or planetary exploration single camera system, a stereo vision system, an inertial missions [1][2] including landers and rovers are earnestly measurement systems, a scan typed laser range finder etc. under studying in Japan. Those missions will follow up The authors developed advanced navigation and

i-SAIRAS 2010 August 29-September 1, 2010, Sapporo, Japan 845 guidance methods including a terrain recognition scheme, speed (more than 12km/s) from an interplanetary a path planning algorithm, a self-positioning method, an earth-return trajectory. HAYABUSA is in operation to intelligent tele-driving system. return back to the earth in June 2010. This paper presents planetary robotic exploration JAXA is also promoting SELENE mission, in which scenario by mobile explorers. Then this paper describes the lunar orbiters were launched in 2007. The main the system configuration of the developed test-bed rovers orbiter, Kaguya has performed the highly accurate and for long traverses and rover-based scientific observation. high-resolution observation of elementary and miner This paper also presents the detailed functions and shows compositions of surface, topography, subsurface the performance of the developed rover test-beds. The structure, magnetic anomaly and gravitational field. effectiveness of the developed smart manipulator is JAXA/ISAS is planning a Venus orbiter mission, shown by experimental results, sample collection, Planet-C, which launch will be scheduled in FY2010. digging etc. The developed rover test-bed is used for Planet-C will elucidate the mechanism of the feasibility study of the future lunar or planetary super-rotation, the circulation of the Venusian exploration missions. The exploration scenario or atmosphere at speed 60 times higher than the rotation strategy is also verified by the developed rover test-bed speed of the planet, and will investigate and identify the through some field tests.. movement of the atmosphere by infrared and other observations and will investigate the distribution of lightning discharges. 2 Planetary Explorations in Japan BepiColombo, a large Mercury exploration mission of ESA, is planned to be implemented as a joint project Since the launch of the first Japanese artificial between ESA and JAXA. ESA will send two explorers to satellite Ohsumi in 1970, ISAS (Institute of Space and Mercury. The Mercury Magnetospheric Orbiter (MMO) Astronautical Science), which is now the space science developed by JAXA will be carried together with the division of Japan Aerospace Exploration Agency (JAXA), Mercury Planetary Orbiter (MPP) of ESA. MMO mainly has been responsible for space science missions of Japan, conducts the magnetic field and magnetosphere from conceptualizing through developing, launching and observations. MMP mainly conducts the surface and operating spacecraft. In the course of the cooperative and internal observations. concerted efforts among engineers, scientists and For the time being, JAXA will proceed with the near industries in developing and operating missions, ISAS future plan focusing on terrestrial planets, including the has made progress in technologies and accumulated Moon and asteroids, keeping Jupiter exploration in view experiences for requirements to achieve the recent and as a future plan. To study the origin and evolution of the the near-future space science missions. solar system, asteroid exploration missions have been The planetary missions, including the lunar one and studied, because asteroids preserve the information from those by planetary landers and rovers require a high the age of the solar system’s birth. Utlilizing engineering degree of autonomy in every aspect of spacecraft technologies obtained from the HAYABUSA explorer, an performance. For such missions, navigation, guidance asteroid multi-sample return mission is under study as a and control are no exception, and the technologies have next small body mission, which includes global been required. Those are intelligently adaptable to observation of the surface layer, investigation of internal unknown or ambiguous environment. With this in mind structure, and collection of surface material, by the and to cope with the launch weight limitation, a variety remote-sensing instruments and the surface exploration of sensors, actuators and control systems have been robots. investigated to develop a small explorer with light As the large scale lunar exploration missions, weight and low power consumption. SELENE series are under study, internal observation, JAXA is earnestly performing, planning and studying geological observation by lander and rovers, lunar lunar or planetary exploration missions in Japan. observatory, sample return missions etc. For Mars MUSES-C spacecraft, renamed HAYABUSA in orbit, missions, the Mars orbiter and Lander-Rover missions meaning falcon in English, was launched on May 9, 2003. are under study. To investigate the internal structure of The primary goal of the mission is to develop and verify Mars, the lander will land at a place where the prior technologies that are necessary to retrieve samples from explorations suggest that a groundwater layer exists with a small body in the solar system, with the obvious but high possibility and rovers carry the artificial earthquake surplus science return of obtaining fragments from the source and a broadband seismometer. A venus balloon asteroid 1998SF36. The major technologies to be mission is also under study. By floating a balloon in the verified include among others 1) use of solar powered deepest region of the Venusian atmosphere, it is possible electrical propulsion system for an interplanetary mission, to conduct meteorological observations, ground surface 2) autonomous navigation, guidance and control to a observations, measurement of atmospheric composition small target in space, 3) rendezvous with an asteroid and and the others. sampling of the surface material, and 4) reentry at a high

846 3 Rover Test-bed Model but also integrated and advanced mission management technology. The Micro6 will install manipulator system, The authors have developed several rover test-beds tool system, scientific instruments for detailed surface by far in series [3]. As so called “Micro5”, the design exploration. The key technology of Micro6 project is to concept when it was in an early stage of the project was develop intelligent software architecture. to realize small enough to be loaded on the piggyback payload of the launch vehicle, and to push all of advanced technologies of mobility, sensing, control, navigation, manipulation and their processing systems into tiny bodies [4][5][6]. Originally, three models of the Micro5s were developed. They are for testing mobility, for manipulation, and for navigation, respectively. The designed rover is driven by five wheels controlled independently. The steering is controlled by the differential of left and right wheels. The designed rover has the new suspension system called "PEntad Grade Assist SUSpension" (PEGASUS) [7]. After research with these three conceptual models of Micro5s, the fourth and fifth models have been developed. Both models are relatively bigger than the first three, to dedicate the research for more advanced Fig. 1. Conceptual Design of Micro6 navigation and tele-science technology. The results of researches with these two models, visual navigation, localization and mapping, and manipulation techniques 4 Micro6 System were developed. The models were used for the basic study of 4.1 Mobility system technology required in surface exploration. However the One of the key issues to make such small rovers integrated systems are required to investigate what practical is the mechanism of mobility. Since even small problems lurked to achieve the missions. Therefore, a obstacles should be relatively big for a small rover, the new integrated test-bed has been designing as a rover mobility system has to have better performance both to system which installs all the elements developed so far. traverse rough terrain and to simple and lightweight As a new test-bed, Micro6 has been projected as a mechanism, simultaneously. The authors have developed sequel to the Micro5, constructing a series of rover a lightweight, simple mobility system for micro rovers prototypes which have capability to carried out a variety known as PEGASUS, five wheeled grade-assisting of the novel mission sequences. Micro6 is not designed suspension mechanism. PEGASUS consists of a for the mission specific, but for pushing the technology conventional four-wheel drive system and a fifth active advance. So the Micro6 has been designed with wheel. The fifth wheel is attached to the end of a link, comprehensive knowledge of past research of the and the other end of the link is attached to the body with Micro5s. Figure 1 shows the conceptual design of a passive rotary joint. The system is designed to Micro6. First, the authors think the mobility system for distribute the load of weight to all five wheels when the the rover systems have to have not only high mobility rover climbs up on the step-alike terrain. but also fault tolerant feature. The total traversability of PEGASUS has remarkable performance when it the rovers must be limited by their trustiness on unknown climbs up hazardous surface. However, it does not circumstances rather than their speed, climbable angle, contribute safety when the rover descends downhill. energy or heat problem. Surface exploration missions Micro6 is equipped with a suspension mechanism so might be dramatically enhanced if the rover could run called HEXUS [8], a six wheeled mobility system with trustingly without unrecoverable failure such as to fall passive / semi-active / active switchable suspension. down into a hidden ditch. The Micro6 has the novel Thanks to the new suspension system, HEXUS has not suspension system called HEXUS which has failure only good performance as well as PEGASUS, but also tolerant feature. Wheel design must be done for mission has great stability and fault tolerant feature. HEXUS oriented, because its performance is seriously affected by could also choose mobility modes depending on the surface condition. On the other hand, the wheel design is environment, so that it could reduce energy consumption one of the most difficult problems because the surface on flat surface. condition is usually unknown before landing. The authors have tested various types of wheels. Secondly, (1) Fast Cruise Mode: Micro6 is still no more than a test-bed rover on the When the vehicle traverses on relatively flat surface, ground, however, targeting not only elements research one of the most dangerous situations might not be

847 collisions to obstacles, but falling into hidden ditches. Micro6 have to have a thicker and heavier structure to The anxious which the vehicle might fall into hidden apply for on 1G environment compare with those for on ditches would make the cruising speed slow. Fast cruise 1/6G. mode of HEXUS would reduce the anxious of the failure. Since the first prototype of Micro6 is almost ten The mode is deploying five of six active wheels as times heavier than the first models of Micro5s. So reversed PEGASUS. The forward center wheel touches Micro6 needs a steering mechanism on the wheel in four to the ground, and the other end of the suspension arm each corner, respectively. Each steering mechanism is connected to this wheel is attached to the body with a modularized with a drive mechanism and a motor control free joint. processor with a motor driver circuitry. For use of bigger (2) High Mobility Mode: wheel diameter, the whole drive/steer mechanisms are placed inside the wheels. Thus, the drive axes are slightly When the vehicle is climbing a hill up continuously, tilted to earn enough angle of steering. This may cause high mobility mode would be good. This mode is just the reduction of mobility performance on softer soil surface. same configuration of PEGASUS, where the center rear wheel is due as the fifth drive wheel. With five wheels 4.3 Wheels Design drive, the vehicle would gain a great mobility to climbing rocky hill up under relatively low energy Wheel is one of the most difficult parts to design, consumption compare with six wheels drive system. because it is impossible to know where or what kind of terrain the vehicle would traverse beforehand. It is (3) High Mobility with FT Mode: required to fulfill a lot of requirements for the When the vehicle traverses on harsh terrain, a fault mechanical design of the mobility system, e.g., to earn tolerant (FT) feature must be essential. HEXUS adds this enough angle of steering, larger wheel diameter is feature by using full six wheels to drive. Though it preferred but lighter weight and strong structure are consumes more energy than PEGASUS does in this needed, etc. The tilted teacup wheels would be suited for mode, it could relieve anxiety. The center forward wheel harder rocky surface rather than softer sandy surface. is used as a mechanical sensor while the center rear However, it is not a big issue at this moment because the wheel gives full play to its mobility. By any chance, if compact set of drive/steer/control modules of Micro6 the forward wheel(s) fell into the hidden ditch, the center enables to use any type of wheels. forward wheel could sense the situation mechanically, and it could support the body of the vehicle to prevent 4.4 Electronic Systems downfall. The vehicle would be able to come back from Micro6 introduces highly modularized architecture to the edge of ditch, simply by moving backward. both hardware and software. Software modules are (4) Active Suspension Mode: constructed with RT middleware or equivalents, could be The active suspension mode is a promising feature to placed on any processing elements connecting via high overcome the limitation of the wheels. The wheels are speed network. Due to the modularized approach, each good for harder smooth surface, but not for soft sandy or software module can be designed to be highly reusable. harsh surface. For these surfaces, crawler or legs might Team production is also available by the network have better performance; however, in terms of energy distributed system. The hardware system is also designed efficiency and simplicity, the wheels would still be better for network distribution. This approach is not optimal for in many cases. The active suspension mode might have energy or weight, but is better way for ease of software advantages of both wheels and legs. Though, the authors development. The first model of Micro6 has at least do not have any evidence to prove that the active more than ten processing units including vision suspension had superb performance yet, it would allow processor. researchers to develop optimal motions against variety of surface conditions. 5 Intelligence for Navigation 4.2 Link and Steering Mechanism In the case of a remote environment such as the Unlike the first models of Micro5s, the Micro6 moon, Mars, or other planets, some time-delay occurs adopts single piece body design. Therefore, Micro6 has a between a master and a slave system due to the distance pair of locker suspension arms and a differential and the limited capacity of communication bandwidth. It mechanism for bearing its body in the middle of the two is thus not easy to compose a closed loop control arms. The differential mechanism has a newly designed structure between the master and the slave system. To backlash free feature with very simple structure. So the navigate and guide the rover to the given destination, two body is equipped with some sensors, e.g., visions, inertial navigation schemes are usually used, tele-operation and measurement unit, and manipulators, backlash free is an autonomous navigation. Conventional tele-driving important feature. According to the structural strength methods use some strange behavior called "Move & analysis, the suspension arms of the first model of Wait" for operation of a rover. In conventional

848 tele-operation, a rover has to wait for commands while data sets. Therefore, the proposed scheme the operator's planning the path. That requires careful compensates waypoints by using the latest consideration and as a consequence, that causes time measurement data which would be more reliable consumption. Moreover, to avoid collision between the than the previous data sets. This is because the waypoint path and obstacles, a rover requests the measurement data of a certain area is more reliable, operator to regenerate the waypoint path, which causes when a rover is getting closer to that area. Here, further delay until a new path data is received. The assume that the difference between those terrain authors have studied autonomy and intelligence for maps is the distortion between data sets, and the navigation of a surface explorer. The authors have path would be compensated by using a distortion studied tele-driving system [9] for efficient and safe compensation matrix which is the mapping between exploration. This paper presents the intelligence for the old and new terrain data sets. tele-driving. The proposed tele-driving system can be available for human operation on the ground as a support 5.2 Autonomous Visual Navigation System system. Autonomous navigation system is also studied. The authors have studied visual navigation system. This A visual sensor is used for navigation and scientific paper presents two schemes, visual navigation by using a observation. Exploration robot can get the global single vision and SLAM scheme by stereo vision. information on the environment from the vision system. Image obtained from a single camera has a lot of 5.1 Intelligent Tele-driving System environment information from the area near rovers to the skyline. However, it is difficult to navigate a robot to the An intelligent tele-driving scheme has been destination with only gray-level images, because the developed for safe, efficient and continuous driving farther area's information is more ambiguous. of the rover, which should be a low-level intelligence Accordingly, this paper proposes a method to know that can understand human intentions in the the situation and plan the behavior based on gray-level operator's path command for obstacle avoidance. images. Firstly, goal information is commanded to a Here, a human machine cooperative tele-drive robot from an operator based on image data. And then an system is discussed, consisting of a global and a local exploration robot recognizes the environment widely path planning for long range traversability. The from gray-level image and then creates a map and plans operator can create any desired command-path as a a path by generating way-points. When a robot detects sequence of waypoints by using a 3D terrain model dangerous-area, a robot can set way-points to avoid obtained as DEM data by an on-board sensor. Those obstacles. data are transmitted to the ground station. A (1) Environment recognition dangerousness map is then built using the received When the gray-level images of the surface of the terrain data. However, the measured terrain model moon or planets are shown, the following recognition is may include some errors and cause some problems. performed. White area is safe area, and black & white For example, generally, data measured by a sensor area is rough area, whose shades change a lot. Hence the has proportional errors depending on the distance safe-areas for rovers are the whiter area and the area from a sensor to a measured target, and unknown whose shades do not change so much. In the gray-level obstacles might be found on the way of the path, image, the shades indicate the rate of white and the because of an occlusion problem of sensors such as a variances indicate the changes of shades. stereo camera. Of course, a rover itself also causes For the extraction of the dangerous-area, the gray-level image is divided into mesh. One of meshes is position estimation errors and dead reckoning errors, called "window". From the result of calculation for the because of slips of wheels etc. For corresponding to average of shades and the variance in each window an unknown obstacle, a conventional autonomous created on the image, dangerous-areas are extracted. path planning algorithm is one of solutions, and it Dangerous-areas are also recognized as landmarks for can be applied for a short range path planning position estimation between each waypoint. (2) Map building On the other hand, a rover is continuously Map on the robot coordinate system is build based on updating the environment data set, and calculates the image data as shown. Look-up-table between the the difference between original terrain data sets image coordinate value and the robot coordinate value used for initial path planning by the operator and is prepared in advance. the data sets acquired by the on-board sensors of the (3) Path planning rover. The original path may indicate the rover to Each window on the created map has the value follow a trajectory that might cause a collision to which indicates the level of danger. When the value is obstacles, due to the difference between the larger that the threshold, then that area is judged to be distorted original and the more accurate acquired obstacle area. The threshold can be decided by the

849 average level of the intensity of the all the pixels on the 7 Conclusions image. The created path from the start point to the goal points includes some information, way-points, the This paper has presented Japanese lunar or planetary direction and distance between way-points. Way-points robotics exploration missions. Especially surface can be defined the window area where a rover will explorers have been also studied for the future lunar or change the moving direction. The path planning planetary missions requiring long traverses and algorithm is proposed as follows. rover-based scientific experiments. This paper proposed At the first step, the rover makes the straight path a new rover architecture. This paper also discussed between the start point WAS and the goal point WAE. At autonomy and intelligence for a navigation and guidance. the second step, the robot searches whether the area on This paper presented an intelligent tele-driving system, a the obtained path includes any obstacle or not. If it visual navigation system and SLAM system by stereo includes, then go to the third step. Otherwise go to the vision. The proposed functions are under development in sixth step. the proposed test-bed rover. At the third step, the candidates of way-points is newly built on the right side or left side of obstacle area to avoid obstacle. At the fourth step, the avoidance References direction will be decided by evaluating the avoidance distance from the path and the level of the danger. And [1] T.Kubota, Y.Kunii, Y.Kuroda, T.Yoshimitsu, T.Okada, then a new way-point is decided as WAi. At the fifth step, M.Kato : Lunar Robotics Exploration by the rover makes the straight path between the current Cooperation with Lander and Micro Rovers , 6th point WAi-1and WA1. Then go to the second step. IAA Int. Conf. on Low-Cost Planetary Missions, At the sixth step, the rover calculates the distance pp.189-194, 2005. and direction data between each WAi (i= S, …., E). [2] T.Satoh, Working Group for MELOS Mars Here the width of the path is the same as the width of Exploration Mission, Mars Exploration Mission the robot. MELOS: An Overview. Japan Geosciences Union Meeting 2009, P142-001, 2009. [3] T.Kubota, Y.Kunii, Y.Kuroda, M.Otsuki, Japanese 6 Test-bed Rovers Rover Test-bed for Lunar Exploration, Int. The authors have developed rover test-beds. Figure 2 Symposium on Artificial Intelligence, Robotics and shows the developed rovers. The specification of the Automation in Space, No.77, 2008. developed test-bed rovers is shown in Table 1. The 5th [4] Y.Kuroda, K.Kondo, K.Nakamura, Y.Kunii, T.Kubota, and 6th wheels are under development and the Low Power Mobility System for Micro Planetary manipulator systems have been development. Rover “Micro5”, Proc. 5th iSARAS, pp.77-82, 1999. [5] T.Kubota, Y.Kuroda, Y.Kunii, I.Nakatani, Micro Planetary Rover "Micro5", Proc. of 5th iSAIRAS, pp.373-378, 1999. [6] T.Kubota, Y.Kuroda, Y.Kunii, I.Nakatani, Lunar Exploration Rover:Micro5, Advanced Robotics, Vol.14, No.5, pp.443-444, 2000. [7] Y.Kuroda, Y.Kunii, T.Kubota, Proposition of Microrover System for Lunar Exploration, Journal of Robotics and Mechatronics, Vol.12., No.3, pp.91-95, 2000. Fig. 2. Developed Micro6 Rover [8] Y.Kuroda, T.Saitoh, Y.Kunii, T.Kubota, System Table 1. Specification.of Micro6 Design of Micro6 Rover for Longer Range Scientific Exploration. Int. Symposium on Artificial Size 1.1[m](W), 1.2[m] (L), 1.1[m] (H) Intelligence, Robotics and Automation in Space, Weight About 50[kg] Mobility System HEXUS, (6WD) No.97, 2008. Wheel diameter : 0.2[m] [9] T.Kubota, M.Otsuki, Y.Kunii, Y.Kuroda, Surface Mobility Velocity : 0.2[m/s], Exploration Rover and Guidance Scheme for Performance Climable step : 0.30[m] Planetary Robotic Exploration, 26th Int. Symposium Climbable slope : 20[deg] on Space Technology and Science, 2008-d-04, 2008. Power Supply Solar Panel, Battery : Lithium Payload Stereo cameras, LRF, SI, IMU Clinometers,Manipulator

850