Optimal Motion Planning of a Space Robot with Base Disturbance Minimization Authors: Eric Kaigom, Thomas Jung, Jürgen Rossmann Institute forman MachineInteraction, RWTH Aachen University 1
Virtualize Robotic Applicationsto spare PhysicalPrototypes Virtualization is a key technology in Modern Computer Centers Electronic Hardware Development Aerodynamics etc. 2
The first Virtual Testbed : The International Space Station 3
The first Virtual Testbed : The International Space Station 4
delivered the key ideas 5
in practical applications! 6
Glueing in a Virtual Testbed for Planetary Landing 7
What is a Virtual Testbed? More than one process type must be simulated within the same virtual world/testbed Inverse kinematics, rigid body simulation, terra mechanics, sensors, mechatronics, hydraulics, pneumatics, aerodynamics, geo data, fire and water propagation, etc. Glueing strategies needed to connect the results Complex environments become part of the simulation models Large world models, multiple interacting agents, streaming, etc. Flat file storage replaced by world model databases Modeling capabilities must keep up with complexity Support of advanced metaphors Computational power increases through multi core processors Testbed must exploit parallelism Multi-threaded rendering Multi-threaded simulation Basic challenge: Design the structure of a world simulator 8
Application -Orbital Servicing usingspace Robots Assembling Inspection 7-DoF Manipulator Space Robot Base Repair Upgrade Satellite Refueling Maintainance De-Orbiting Capture 9
Motivation Newton s Law in Space 10
Motion Planning -Reactionless Joint Velocity Conditionfor reactionlessjointvelocity Space robotat rest at the beginningandat the end ofits motion Noexternal forcesandmomentumsactingon thespace robot Reactionless velocity profiles givenas: The reactionlessvelocityprofilesdepend on thecurrentdynamic propertiesofthe spacerobot,because: 11
Motion Planning -Feasible Trajectories at Velocity Level Reactionlessvelocity: Null spacebase: Feasibletrajectories: 12
Motion Planning -Trajectory Parameterisation Polynomialtime-varyinginputs: Boundaryconditionsatthe startandtheend of the motion: The motionplanningproblem: Find is reachedat the end ofthemotion such that thedesired TCP-pose 13
The Constrained Particle Swarm Optimization (C-PSO) Noneedof gradient-information, only acostfunctionisrequired The optimizationvectorsexplore thesolutionspace usingthe followingscheme: ConstrainedPSO: Only those optimizationvectorsrespectingthe jointpositionbounds areretainedaspossible solutionofthe motionplanning problem 14
Simulation Results Approach without Base Disturbance 15
Technology Transfer: Virtual Testbedfor ProductionEngineering 16
Interaction between Aerodynamics, Water and Fire 17
A Virtual Testbed in the woods : Driver training 18
Virtual Testbed: Bulk solids simulation 19
Interactive tests in a Virtual Testbed 20
Metaphors tovisualize motor torques and contact forces 21
A Virtual Testbed for thephysicalplanetary LandingMockup 22
Space technology is themotor for terrestrialvirtual Testbeds Virtual Production Virtual Forest Working Machines Projective Virtual Reality Geovisualization Virtual Construction Site 23
Conclusion The proposedalgorithm: Exploitthe ReactionNull Space andpolynomialtime-varying inputstosteerthe manipulatorofa spacerobottoa targetpose The base disturbanceduringthe approachwas minimized The C-PSO delivers theoptimizedjointtrajectories The virtualization approachissuccesful: Virtual Testbeds are a key concept to face the challenges of research and engineering tasks in complex environments. Important process simulation components are already integrated (actuator dynamics, sensors, control, co-simulation, etc. ) Optimization techniques can readily be applied to the virtualized robotic applications and thus spares physical prototypes 24