NASA RASPBERRY SI

https://nasa-raspberry-si.github.io/raspberry-si/

Transcript

1. RASPBERRY SI David Garlan CMU Co-I Bradley Schmerl CMU Co-I Pooyan Jamshidi USC PI Javier Camara York Collaborator Ellen Czaplinski Arkansas Consultant Katherine Dzurilla Arkansas Consultant Jianhai Su USC Graduate Student Matt DeMinico NASA Co-I AISR: Autonomous Robotics Research for Ocean Worlds (ARROW) Resource Adaptive Software Purpose-Built for Extraordinary Robotic Research Yields - Science Instruments Abir Hossen USC Graduate Student
2. K. MICHAEL DALAL Team Lead USSAMA NAAL Software Engineer LANSSIE MA Software Engineer Autonomy • Quantitative Planning • AI-based Mission Discovery • Transfer & Online Learning • Model Compression JIANHAI SU USC, Graduate Student BRADLEY SCHMERL CMU, Co-I DAVID GARLAN CMU, Co-I JAVIER CAMARA York, Collaborator MATT DeMINICO NASA, Co-I HARI D NAYAR Team Lead ANNA E BOETTCHER Robotics System Engineer ASHISH GOEL Research Technologist ANJAN CHAKRABARTY Software Engineer CHETAN KULKARNI Prognostics Researcher THOMAS STUCKY Software Engineer TERENCE WELSH Software Engineer CHRISTOPHER LIM Robotics Software Engineer JACEK SAWONIEWICZ Robotics System Engineer ABIR HOSSEN USC, Graduate Student ELLEN CZAPLINSKI Arkansas, Consultant KATHERINE DZURILLA Arkansas, Consultant POOYAN JAMSHIDI USC, PI RASPBERRY SI Physical Testbed Virtual Testbed AISR: Autonomous Robotics Research for Ocean Worlds (ARROW) CAROLYN R. MERCER Program Manager Develop Develop and maintain Evaluate Evaluate Develop and maintain
3. Science Discovery in Remote Planets The Phoenix Mars Lander Image Credit: NASA/JPL
4. Current Practice of Science Discovery in Remote Planets Problem: Slow science discovery due to lack of full autonomy 4 (v) Only deal with: Known Knowns ~2.5 hours ~2.5 hours (iv) Does not scale (i) Delay in science discovery (iii) High risks (ii) High mission costs
5. Ideal Vision of Science Discovery in Remote Planets Solution: Fast science discovery with AI-based full autonomy 5 (v) Can deal with: Unknown Unknowns High Frequency Low Frequency (i) Fast science discovery (ii) Low mission costs (iv) Does scale (iii) Low risks
6. Europa Lander Mission Image Credit: NASA/JPL 6
7. Ocean worlds are the best places to search for life. 7 Image Credit: NASA
8. 8 Image Credit: NASA Europa’s surface has many features of interest for a science mission.
9. 9 Image Credit: NASA/JPL-Caltech
10. Autonomy and Robotics Image Credit: EuroScience 10
11. Autonomy Module: Design 11 • MAPE-K Loop based design • Machine learning driven quantitative planning and adaptation Monitor Analyze Plan Execute Knowledge System Under Test (NASA Lander) Autonomy
12. Autonomy Module: Planner 12 PRISM Model Generator Policy Generator (PRISM) Plan Extractor Plan Translator Run-time Information PLEXIL Plan High-level Plan Policy File PRISM Model Planning Monitoring and and Analysis Excavation Locations Position (X, Y) Science Value Excavation Probability xloc1 (1.93, 0.2) 0.37 0.85 xloc2 (1.51, -0.4) 0.41 0.83 xloc3 (1.92, -0.3) 0.42 0.83 … Dump Locations Position (X,Y) dloc1 (1.46, -0.9) dloc2 (1.49, 0.6) [select_xloc3, select_dloc2] Excavation: { Boolean planSuccess; LibraryCall GuardedMove ( X = 1.92, Y = -0.3, Z = 0.05, DirX = 0, DirY = 0, DirZ = 1, SearchDistance = 0.25); planSuccess=Lookup(GroundFound); if (planSuccess) { LibraryCall Grind X = 1.92, Y = -0.3, Depth = 0.05, Length = 0.2, Parallel = true, GroundPos = Lookup(GroundPosition)); planSuccess=Lookup(DiggingSuccess(0.83)); if (planSuccess) { LibraryCall DigCircular ( X = 1.92, Y = -0.3, Depth = 0.05, GroundPos = Lookup(GroundPosition), Parallel = true); LibraryCall Deliver ( X = 1.49, Y = 0.6, Z = 0.5); } endif } endif } Execution
13. Autonomy Module: Evaluation 13 Design • MAPE-K loop based design • Machine learning driven quantitative planning and adaptation Evaluation • Two testbeds: different fidelities & simulation flexibilities Monitor Analyze Plan Execute Knowledge System Under Test (NASA Lander) Autonomy Physical Testbed OWLAT (NASA/JPL) Virtual Testbed OceanWATERS (NASA/ARC)
14. Physical Testbed at NASA/JPL Ocean World Lander Autonomy Testbed (OWLAT) 14 Physical Autonomy Testbed: https://www1.grc.nasa.gov/wp-content/uploads/2020_ASCE_OWLAT_20191028.pdf E2M Technologies six DOF Stewart Platform representing spacecraft lander Barrett WAM seven DOF arm mounted to lander with wrist FTS and tool changer Modular instruments to be mounted on robot arm Testbed setup and major components HITL simulator of lander and manipulator
15. Virtual Testbed OceanWATERS by NASA ARC 15 Surface Model Lander Model Surface/Arm Interaction Material Property Visual Appearance Lighting Model Celestial Bodies Morphology Sensor Data: Galileo Orbit 26, Europa Observation Fractal Terrain Generation • Analytical scoop model implemented for real time approximation • Particle simulation modeling: Discrete Element Method • Actuated Arm • Kinematic models • Dynamics and collision handling • Antenna: pan-tilt unit • Power and battery modeling • Tools: camera, scoop, grind Source of images: An Autonomy Software Testbed Simulation for Ocean Worlds Missions. Source Code of OceanWATERS Testbed: https://github.com/nasa/ow_simulator
17. Homepage: https://nasa-raspberry-si.github.io/raspberry-si