Autonomous Sample Acquisition for the ExoMars Rover

Size: px

Start display at page:

Download "Autonomous Sample Acquisition for the ExoMars Rover"

Daniela Jefferson
5 years ago
Views:

1 In Proceedings of the 9th ESA Workshop on Advanced Space Technologies for Robotics and Automation 'ASTRA 2006' ESTEC, Noordwijk, The Netherlands, November 28-30, 2006 Autonomous Sample Acquisition for the ExoMars Rover Dave Barnes (1), Andy Shaw (1), Steve Pugh (1), (1) Department of Computer Science, University of Wales, Aberystwyth, SY23 3DB, UK. ABSTRACT The Mars Exploration Rovers (MER) have achieved a 10 fold increase in traverse distance travelled per Sol when compared with the first rover to land on Mars, and are now in a further extended mission phase. This success is not without cost as primary mission operations can often cost as much as 40% of the total primary mission. The fundamental problem is that large teams of scientists and engineers are involved in the tasks of defining, rehearsing, planning, scheduling and uploading every single activity associated with surface operations, no matter how small or large that activity might be. The ESA ExoMars rover is planned to traverse further and for longer than the MER rovers, and therefore these operations problems will arise. If they are not solved, then potential science data will be lost and operations costs will soar. The ability for a rover to operate autonomously is advantageous as this could potentially increase science data return whilst reducing operations costs. Surface science sample acquisition is a critical activity within any planetary exploration mission, and our research is focused upon the design, implementation, experimentation and demonstration of an autonomous surface sample acquisition capability for the future ESA ExoMars rover mission. This paper presents our recent work in this area. INTRODUCTION Current work in the area of autonomous science data gathering for Earth observation satellite operations is showing how successful an autonomous approach can be [1]. However to apply such an approach to planetary surface rovers is not trivial, and JPL are investing in this area [2]. Past [3] and current [4] Mars rover exploration has shown that intensive Earth-based operations are required to support such missions, and the time delays that can accumulate whilst even the simplest of activities is being planned for the operations time-line. From the current MER mission it has been found that once a science target has been identified by ground based planetary scientists, it still takes up to three sols for a rover to acquire the planned science sample [5]. To increase science return, this acquisition time needs to be reduced to less than one sol. Our research is focused upon the problem of autonomous sample acquisition, and in the first instance we have focused upon surface sampling. Our initial work has used past data obtained from the rehearsals work that was undertaken when we were preparing for the 2003 Beagle 2 mission [6]. A virtual Beagle 2 simulation model was developed and calibrated with the flight Beagle 2 hardware, namely, the 5 DoF robotic arm and the PAW Stereo Camera System. As this simulation contained a calibrated arm kinematics model, it could be used to generate the arm joint parameters that would place the arm in a desired configuration after each joint had been commanded in turn to rotate by the designated number of degrees (see figs. 1, and 2). Fig. 2 shows a screen shot of our simulation model to the top-right of the image. In addition to instrument placement rehearsals we also spent considerable time generating digital elevation models (DEMs) of the rock garden that had been assembled within the Beagle 2 Lander Operations Centre (LOCC) as part of the Beagle 2 Ground Test Model (GTM). It is this past DEM data that we have been using for our initial autonomous surface sample acquisition experiments. In addition to our Beagle 2 work, we were subsequently involved with the EADS Astrium led ESA ExoMars Phase A study [7]. ExoMars rover science operations simulations were developed and we plan to use this work This work is being funded under the UK PPARC CREST Grant No. PP/D00666X/1, and is in collaboration with SciSys Ltd., the University of Leicester and the University of Strathclyde.

Fig. 1. The Beagle 2 PLUTO Mole undergoing operations rehearsals at the Lander Operations Control Centre. Fig. 2. A PLUTO Mole acquired sub-surface sample being placed within the Gas Analysis Package (GAP) Inlet Port using the Beagle 2 ARM.

3, and 4 show stills from our ExoMars simulation/animation work and the Beagle 2 ARM with Rock/Corer Grinder (RCG) can be clearly seen.

2 Fig. 1. The Beagle 2 PLUTO Mole undergoing operations rehearsals at the Lander Operations Control Centre. Fig. 2. A PLUTO Mole acquired sub-surface sample being placed within the Gas Analysis Package (GAP) Inlet Port using the Beagle 2 ARM. as part of our autonomous surface sample acquisition experiments. Figs. 3, and 4 show stills from our ExoMars simulation/animation work and the Beagle 2 ARM with Rock/Corer Grinder (RCG) can be clearly seen. In addition to Beagle 2 and ExoMars rover simulation work, our experimentation and demonstration will utilise the UWA Planetary Analogue Terrain (PAT) Laboratory ( Mars Yard ) and our new Concept-E rover chassis and mounted manipulator arm. We define a sample as both instrument data (e.g. from an X-ray spectrometer), and real physical material (e.g. a rock core or surface/sub-surface soil). Before autonomous sample acquisition can occur, one must initially determine a suitable instrument placement site, and our work in this area is first described. AUTONOMOUS INSTRUMENT PLACEMENT SITE SELECTION Fig. 5 shows the Beagle 2 GTM with the 1/3 mass PAW mounted onto the robotic ARM. This PAW was a rapid prototyped version of the flight PAW and due to its mass it generated ARM deflections comparable to those that would have been generated under Martian gravity. The 1/3 mass PAW had two COTS cameras positioned so as to emulate the flight Stereo Camera System (SCS). Image pairs captured using these cameras were processed to generate a 3D model of objects within their field of view. Adjacent to the Beagle 2 GTM a Mars terrain mock-up was constructed, see fig. 6, and by moving the ARM with the 1/3 mass PAW over this terrain, a mosaic of camera image pairs could be captured and a DEM of the terrain mock-up generated, see fig. 7. A DEM such this, has formed the starting point for our autonomous instrument placement site selection work. Using a terrain DEM as input, we have developed a rock feature extraction algorithm which is based upon the radius of curvature of neighbouring points upon a rock s DEM surface. Six features are currently characterised, and are colour coded for ease of visualisaton (Red = Peak, Magenta = Ridge, Green = Pass, Yellow = Plane, Blue = Channel and Black = Pit). Fig. 8 shows the colour coded feature map for the test DEM. The percentage of extracted features from a scene can be varied by adjusting the slope, minimum curvature and resolution thresholds. The slope threshold determines an angle, below which the terrain is said to be horizontal. The minimum curvature threshold is an angle above which the terrain is said to be curved, and the resolution threshold determines the number of surrounding data points used in the calculations. Given the fractal nature of rocks, a large number of features can be generated, but by automatically adjusting the slope, curvature and resolution threshold values, this number can be reduced as and when appropriate. Using the feature extraction parameters, our initial experiments have focused upon planar rock regions that are close to zero degrees with respect to the horizontal. We argue that such regions are a good place to start (from an engineering stand-point) when searching for instrument placement sites. Planar regions that have an 45 degree slope with respect to the horizontal could constitute other placement site candidates. Clearly, such decisions depend upon the the kinematics model of the robot arm and the geometry of the attached instrument 2 2

Collisions between an adjacent instrument and the rock surface under investigation must be avoided at all costs.

3 Fig. 3. Image of the ESA ExoMars rover. EADS Astrium led Phase A Study team. Note the Beagle 2 heritage robotic ARM with Rock/Corer Grinder RCG). Fig. 4. Image of the ESA ExoMars rover conducting surface science with the aid of its robotic ARM. EADS Astrium led Phase A Study team. package. Collisions between an adjacent instrument and the rock surface under investigation must be avoided at all costs. Nevertheless, our feature extraction approach is sufficiently versatile to enable more complex rock placement sites to be identified than just the simple planar region described here. Once candidate placement sites have been identified, standard image processing techniques are employed to first remove, for example, the planar regions that correspond to the ground surrounding the rocks (assuming that soil sampling is not required). Next each region is searched to identify the largest contiguous region, and finally the centre of area of the resultant region is determined, see fig. 9. The final part of our autonomous instrument placement site selection method involves determining the DEM polygon region that corresponds to the Cartesian position that has been generated from the previous centre of area operation. Once this polygon region is identified, it is then a trivial task to generate a surface normal for this region. A surface normal tag point is also generated for the instrument that is to be placed against the rock surface, see fig. 10. Given both instrument and rock surface normals, a polynomial trajectory between both normals is generated (subject to collision avoidance constraints), and thus a robot arm motion path for sample acquisition is created. The JPL Mechanical and Robotic Technologies Group [9], working with the JPL Multimission Image Processing Laboratory (MIPL) [8] have created a number of image processing software products that generate surface normal and Instrument Deployment Device (IDD) reachabilty maps for the NASA Mars Exploration rovers Spirit and Opportunity. Whilst they too identify flat planar regions, their approach (at present) does not identify any additional 3D rock surface features. Given both the proposed ExoMars PanCam instrument with its High Resolution Camera (HRC) and an arm deployed CLose-UP Imager (CLUPI), we argue that from the stand-point of the planetary scientist searching for evidence of exobiology, the autonomous sampling of rock micro-features may well prove decisive. A UNIFIED KINEMATICS MODEL FOR SAMPLE ACQUISITION As part of our autonomous sample acquisition work, we have recognised that it would be pointless to identify ideal candidate instrument placement sites if insufficient degrees of freedom were possessed by the robotic arm to get to such a site. Due to mass constraints, no mission to Mars has yet flown with a robotic arm possessing six degrees of freedom. Should the planetary science dictate that samples are required from those hard to get to places, then we have been investigating how a unified rover chassis and robotic arm kinematics model might 3 3

4 Fig. 5. Mars terrain mock-up for mission operation rehearsals for Beagle 2. Note 1/3 mass PAW with cameras for DEM generation trials. Fig. 6. Close up of Mars terrain mock-up for mission operation rehearsals for Beagle 2. Fig. 7. Terrain DEM created using Beagle 2 PAW mounted Stereo Camera System. Fig. 8. Colour-coded DEM terrain feature map. 4 4

5 Fig. 9. Processed DEM terrain feature map showing those rock regions above ground level that are flat with respect to the horizontal. These rock regions have been processed to determine the best location for instrument placement - see cross-hair. Fig. 10. Original DEM with generated coordinate frame on the selected polygon surface. A tag point is also present on the instrument (in this example, the Mössbauer Spectrometer), thus allowing an instrument placement trajectory to be calculated. prove useful. The ExoMars rover chassis will require independently motorised wheel levers so that the chassis can be raised off the EDL platform prior to egress. These wheel levers provide the mechanical connectivity between each wheel and the rover body. By moving the wheel levers into appropriate configurations, the rover chassis can be made to roll and pitch, with the wheel steering providing yaw motion. Add to these chassis degrees of freedom a robotic arm with only four degrees of freedom, then a redundant kinematics device is possible. Clearly there are many issues such as transformation commutability, motion range, mechanical collisions etc. to take into account, but we have begun a series of experiments into the formulation of a unified chassis-arm kinematics model for sample acquisition. So far our work has been conducted in simulation using the so-called ExoMars Concept-E rover chassis model. To this chassis model we have added a drill box housing which possesses a three degrees of freedom deployment mechanism. Using our simulation, we have begun to create a unified kinematics model and a number of experiments have been performed. Fig. 11 shows how the model can be used to provide a level drill platform when the rover chassis is positioned on uneven terrain. Fig. 12 shows how the chassis can be pitched so that an instrument mounted on the drill box housing, for example, can be positioned low against a rock surface. It also shows how the chassis can be braced against the soil surface to oppose reaction force induced motion when drilling. Finally, figs. 13, and 14 show how chassis pitch and roll have been added so that instrument placement can be achieved when faced with important hard to get to science sites. PAT LABORATORY TRIALS Whilst our simulation work has proved invaluable for both the autonomous instrument placement site selection and sample acquisition work we have conducted so far, there is no substitute for experiments with real hardware. Accordingly we are in the process of creating a new indoor Planetary Analogue Terrain (PAT) Laboratory. Building work is complete and we are awaiting the arrival of a new eighteen degrees of freedom Concept-E rover chassis (6 wheel driving, plus 6 wheel steering, plus 6 wheel-walking - this provides the independent wheel lever motion), see fig. 15 and fig. 16. Our terrain has been sculpted and has a mole hole for sub-surface sampling experiments, see fig. 17. The terrain will be covered with 15 cm of DLR Mars Soil Simulant-D, together with both rocks acting as obstacles and for corer/grinding purposes, and a sample of science rocks that have undergone prior laboratory analysis. This reference data will allow us to access the accuracy etc. of any rock scientific measurements that have been obtained using our PAT Lab. rover. We intend to modify our robotic arm (see fig. 18) so that it can be mounted on our new chassis together with a mast with pan and tilt mechanism and mounted Wide Angle Cameras (WAC) and High Resolution (zoom) Camera (HRC). 5 5

drill box contact science instrument against rock surface.

Rover chassis with pitch and roll added for accurate

6 Fig. 11. Levelled rover chassis for drill deployment on uneven terrain. Fig. 12. Rover chassis with large pitch for accurate low placement of drill box contact science instrument against rock surface. Fig. 13. Rover chassis with pitch and roll added for accurate placement of drill box contact science instrument against rock surface - View 1. Fig. 14. Rover chassis with pitch and roll added for accurate placement of drill box contact science instrument against rock surface - View

The sculpted Planetary Analogue Terrain awaiting

Obstacle and science rocks will be added once the

7 Fig. 15. CAD drawing of the UWA Concept-E rover chassis in locomotion configuration. Fig. 16. CAD drawing of the UWA Concept-E rover chassis in stowed configuration. Fig. 17. The sculpted Planetary Analogue Terrain awaiting the arrival of the DLR Mars Soil Simulant. Obstacle and science rocks will be added once the soil is in place. Fig. 18. The Beagle 2 PLUTO Mole being tested on the UWA robotic ARM. 7 7

8 CONCLUSION We have begun a challenging research programme aimed at the development of an autonomous sample acquisition capability for the ExoMars rover. With a launch date of 2013, much has still to be done, however we have begun to generate results in the areas of autonomous instrument placement site selection, and the creation of a unified chassis/arm kinematics model to aid sample acquisition. The majority of our work to date has been conducted in simulation. Whilst this has been invaluable, we are looking forward to repeating our experiments with real rover and instrument hardware in our new PAT Laboratory. We have been in discussion with members of the ESA ESTEC Robotics and Automation Laboratory, and we hope to be able to develop a common computer hardware and low level software infrastructure for both our new rover and their ExoMaDeR vehicle. The aim is to be able to run developed software algorithms on both vehicles at both site. This would create an excellent validation environment for any developed rover software. We will report on the results of our PAT Lab. work in the future literature. References [1] S. Chien, et al, Using Autonomy Flight Software to Improve Science Return on Earth Observing One, Journal of Aerospace Computing, Information, and Communication, April [2] T. Estlin, et al, Opportunistic Science with a Rover Traverse Science Data Analysis System, Proceedings 8th International Symposium on Artificial Intelligence, Robotics and Automation in Space (i-sairas), CD-ROM Proceedings, Munich [3] H. W. Stone, Mars Pathfinder Microrover a Small, Low-cost, Low-power Spacecraft, Jet Propulsion Laboratory, [4] S. W. Squyres, et al, The Spirit Rover s Athena Science Investigation at Gusev Crater, Mars, Science, 305: pp , August [5] T. Huntsberger, Y. Cheng, A. Stroupe, H. Aghazarian, Closed Loop Control for Autonomous Approach and Placement of Science Instruments by Planetary Rovers, IEEE/RSJ International Conference on Intelligent Robots & Systems (IROS 2005), Alberta, Canada, August 2-6, [6] D. P. Barnes, N. Phillips, G. Paar, Beagle 2 simulation and calibration for ground segment operations, Proc. 7th Int. Symposium on Artificial Intelligence, Robotics and Automation in Space, isairas 03, NARA Japan, CD-ROM Proceedings, May [7] Dave Barnes et al, The ExoMars rover and Pasteur payload Phase A study: an approach to experimental astrobiology, International Journal of Astrobiology, Cambridge University Press, page 1 of 21, doi: /s [8] A. Douglass et al, Processing of Mars Exploration Rover imagery for science and operations planning, Journal of Geophysical Research, Vol. 111, E02S02, doi: /2005je002462, [9] E. T. Baumgartner, R. G. Bonitz, J. P. Melko, L. R. Shiraishi, and P. C. Leger, The Mars Exploration Rover Instrument Positioning System, Proc IEEE Aerospace Conference, Big Sky, MT, March

NEXT Exploration Science and Technology Mission. Relevance for Lunar Exploration

NEXT Exploration Science and Technology Mission Relevance for Lunar Exploration Alain Pradier & the NEXT mission team ILEWG Meeting, 23 rd September 2007, Sorrento AURORA PROGRAMME Ministerial Council