Testing a Robotic System for Collecting and Transferring Samples on Mars - The Mars Surface Sample Transfer & Manipulation (MSSTM) Project Elie Allouis Elie.Allouis@astrium.eads.net Tony Jorden, Nildeep Patel, Jo Smith and Study Team (see last slide) ASTRA 2011 13 April 2011 - ESTEC
Contents Study and Scenario 1. Introduction and Objective 2. Team 3. Requirements 4. MSSTM baseline Scenario Summary Video Breadboarding and Tests 5. Sample capping/uncapping & discharge. 6. End-Effector 7. Robotic arm control - simulations & vision-control tests Summary 8. Conclusions & Recommendations -2
Introduction and objectives Study and Scope Started in 2009 to investigates the underlying generic technologies for collection and transfer of samples on Mars Sample Acquisition Sample Deposition in Vessel Sample Capping Sample Container Container Pickup Transfer to MAV MAV Launch Sample Protection Requirements The SSHS shall not cause physical changes (e.g. breakdown, mixing, compaction) to the sample. The SSHS shall not cause sample cross-contamination exceeding 1% in mass..while minimising the mass and power requirements, compatible with the Martian environment and being fully compliant with Planetary Protection procedures. -3
Introduction and objectives Phase 1: Comparison and critical review of available options, including a mobile option for sample collection. Develop preliminary designs for Mars surface sample handling and selection of bread-boards. Phase 2: Develop and test breadboards of critical technologies to reduce the risk and raise TRL level for later implementation. Identify technology areas that will need further development -4
Introduction - the Team -5
The MSSTM Baseline Scenario End-to End Sample Return Scenario Summary Video Illustrates Phase 1 Design decisions Introduces Phase 2 Breadboarding -6
Contents Study and Scenario 1. Introduction and Objective 2. Team 3. Requirements 4. MSSTM baseline Scenario Summary Video Breadboarding and Tests 5. Sample capping/uncapping & discharge. 6. End-Effector 7. Robotic arm control - simulations & vision-control tests Summary 8. Conclusions & Recommendations -7
Baseline Concept The baseline concepts selected for breadboarding are therefore; Integrated SV/SC for overall mass saving Dedicated capping mechanism for robustness of the capping operation (optimised system) Bayonet End-Effector to provide robust alignment and locking Robotic Arm Vision Control to provide a robust arm control for both static and mobile scenario -8
Baseline Concept The baseline concepts selected for breadboarding are targeting generic MSR technologies with special emphasis on critical interfaces MSSTM study Future Activity -9
Sample Packaging Breadboard - Setup Sample Collection and Vessel Capping Test Setup Reliable Sample transfer Reliable Capping Demonstrated at various Sample container location Drill Tool Capping Mechanism Sample Vessels Sample Container carousel -10
Sample Packaging Breadboard Testing Capping/uncapping Function Torques/forces Sample discharge Solid sample/loose sample Sample extraction from SV SV sealing (First level test as not part of study scope) Alignment Capping mech - SV Drill - SV Goal Cap/uncap reliable <15N engage, 1.5N uncap, 3Nm max seal Reliable collection from drill Measure required force No leakage Measure tolerance Measure tolerance Results cap/uncap reliable > 30 times <15N, 1.5N, 2Nm solid sample ok > 30 times loose sample: some lost X 380N force required (0.05mm) Leakage after 1 hour X +/- 6mm, +/- 2 deg +/- 3mm, +/- 8 deg SV cap SV body Measure tolerance +/- 3mm, +/- 1 deg -11
Sample packaging Results Summary Results: All nominal tests have been successful Alignment capabilities are better than foreseen; particularly, the Capping Mech and SV cap interface has good self-alignment capability. Motors are adequately sized, and performed without any problem. Sample extraction can be performed without any damage to the (solid) sample. Lessons learned: Improvement identified to the design of the SV and/or drill tool will be required to collect and deposit unconsolidated material effectively partly due to available drill system. A dedicated BB study is required to develop the sealing capability of the SV design. -12
End-effector breadboard Bayonet Design End-Effector Robust alignment with container Interfaces No unintended release Reliable grasping and locking operations End-effector Sample-container interface unit -13
Verify self alignment capability Lateral Angular Combined Locking Motor currents Cycle tests (wear, stable power) Tightening Torque Motor currents Capacity Thermal Locking & tightening Required heater [eg 10W] Dust effects Alignment Tightening End-effector Breadboard Testing Goal >5mm >5 deg 5 deg/5mm 0.5A max 35 cycles min 4A max. 40 NM min -55C to +20 C Assess Check Check 6 mm min. 10 deg 5 deg/5mm 0.09/0.11A max > 35 cycles 0.7A > 40 NM OK Results no failure ~2.8kJ/degC hex-key jammed x -14
End effector: Results Summary Results The bayonet-catch end-effector performed well, Good ability to mate with the sample-container interface Deals well with misalignment Locking is reliable, after life-tests (>35 cycles) Lessons Learned The hex key (for tightening at the SC interface) requires refinement to prevent jamming with heavy dust. Motors require significant extra power to lock / tighten in the presence of dust. Characterised Sensors (position/status) design need refinement. The hex-key showed some sign of deformation alternative concepts identified. Tightening torque for the SC interface is a critical parameter to be defined at the SC detailed design phase. -15
Robotic Arm Control - Simulations Arm Simulations 20Sim modelling tool and Siconos simulation platform for non-smooth dynamical systems Kinematics and dynamics (including modelling of flexibility) and environmental effects (temperature, illumination...) were included. Forces/torques and power budgets were calculated. Vision Control Robustness Tests under various simulated illumination: Robust Algorithms Marker geometry: Shown to be critical: 2mm error in geometry resulted in ~10mm error in arm position. Overall Simulation Results Very good control accuracy and repeatability capabilities: From a rough <9mm with Look and Move To a more precise 0.8mm with visual servoing, -16
Robotic Arm Control - Vision Testbed Test Platform: The Eurobot test-bed robotic arm was used for vision based tests Test Setup A simulated sample-container is created by using visual targets in view of the camera on the arm. Tests performed to: Characterise the robustness of the target detection as illumination changed or as targets were partially obscured, e.g. from shadows. Measure positioning accuracy and repeatability. Do look-and-move and visual servoing control tests. -17
Robotic Arm Control - Vision test results The hardware tests aimed to verify that the simulations gave representative results: Targets at positions shifted from expected positions (e.g. by e.g. 2 cm and 20 degrees) Visual-servoing was used to see how well the test arm could be positioned. Mean errors of < 0.5mm (S.D= 0.1) and < 0.5mrad (S.D=0.5) were measured with good repeatability. Control scheme comparison Look-and-move : performing consistently worse than visual servoing I,e, 8mm and 43 mrad), as identified during simulations. Better accuracy could have been obtained if further moving steps were applied close to visual servoing Robustness tests Occlusion tests with actual occluded test targets Good positioning obtained even under degraded conditions of illumination. -18
Conclusion Concepts and Architecture: Concepts and technologies have been identified at each stage of the sample collection and transfer Preliminary design of the end-to-end sample handling chain has been performed Breadboard and Tests: Breadboards of critical elements have been built and tested Validity of the designs has been assessed Valuable insights have been gained on specific aspects that could be refined in future detailed designs. E.g. Sample deposition, Vessel Cap or Hex-key design Conclusions The solutions developed and tested here provide a number of valuable key technologies for Sample Return missions These are applicable to a wide range of sample handling and return mission scenarios whether static, mobile, on Mars or beyond. -19
Study team Mars Surface Sample Transfer & Manipulation MSSTM Final Report PART A Astrium: Tony Jorden, Elie Allouis, Nildeep Patel, Joe Smith. Selex-Galileo: Samuel Senese, Rolando Gelmi, Piergiovanni Magnani Ruag: Rudolf Spörri, Beat Zahnd, Tobias Welge- Lϋssen Trasys: Konstantinos Kapellos, Roger Pissard- Gibollet, Tecnomare: Roman Chomicz, Roberto Ferrario EPFL: Reto Wiesendanger