Mobile Robotics and Olfaction Lab,
AASS, Örebro University
# 1 RSS 2015 July 16th 2015, Rome, Italy
Robert Krug Todor Stoyanov Achim J. Lilienthal [email protected] [email protected] [email protected]
# 2 © Robert Krug et al. @RSS – Grasp Envelopes for Constraint-based Robot Motion Planning and Control (Jul 16, 2015)
Problem: relatively simple but fast autonomous pick & place and/or
manipulation
EU-FP7 RobLog: [Krug et al., ICRA, 2013], [Stoyanov et al., RAM, submitted]
Amazon Picking Challenge: winning team RBO
Applications: e. g. order picking or unloading in logistics scenarios
# 3 © Robert Krug et al. @RSS – Grasp Envelopes for Constraint-based Robot Motion Planning and Control (Jul 16, 2015)
Agenda
1. Grasp Synthesis- and Motion Planning
2. Exploiting Redundancy
3. Robust Grasp Execution
4. Preliminary Results
5. Discussion
# 4 © Robert Krug et al. @RSS – Grasp Envelopes for Constraint-based Robot Motion Planning and Control (Jul 16, 2015)
Grasp Synthesis-
and Motion Planning
1
# 5 © Robert Krug et al. @RSS – Grasp Envelopes for Constraint-based Robot Motion Planning and Control (Jul 16, 2015)
1.
Grasp Synthesis- and Motion Planning
Where (synthesis) and how (motion planning) to grasp the target object?
Common solution: Sense-plan-act architecture; sampling-based planning
Grasp Synthesis: offline pre-computation
# 6 © Robert Krug et al. @RSS – Grasp Envelopes for Constraint-based Robot Motion Planning and Control (Jul 16, 2015)
1.
Grasp Synthesis- and Motion Planning
Where (synthesis) and how (motion planning) to grasp the target object?
Common solution: Sense-plan-act architecture; sampling-based planning
Motion Planning: online with feasible – first execution [Berenson et al., Hummanoids, 2007], [Krug et al., ICRA, 2013]
# 7 © Robert Krug et al. @RSS – Grasp Envelopes for Constraint-based Robot Motion Planning and Control (Jul 16, 2015)
1.
Grasp Synthesis- and Motion Planning
Where (synthesis) and how (motion planning) to grasp the target object?
Common solution: Sense-plan-act architecture; sampling-based planning
# 8 © Robert Krug et al. @RSS – Grasp Envelopes for Constraint-based Robot Motion Planning and Control (Jul 16, 2015)
1.
Grasp Synthesis- and Motion Planning
Where (synthesis) and how (motion planning) to grasp the target object?
Common solution: Sense-plan-act architecture; sampling-based planning
Problems:
• Slow due to many futile motion planning attempts
• Difficult to incorporate prior knowledge
• Unnatural and/or sub-optimal trajectories
• Global but with probabilistic completeness
• Ill suited to incorporate contact events
• Ill suited to exploit redundancy
# 9 © Robert Krug et al. @RSS – Grasp Envelopes for Constraint-based Robot Motion Planning and Control (Jul 16, 2015)
Exploiting Redundancy
2
# 10 © Robert Krug et al. @RSS – Grasp Envelopes for Constraint-based Robot Motion Planning and Control (Jul 16, 2015)
2.
Exploiting Redundancy
Where (synthesis) and how (motion planning) to grasp the target object?
Central Idea: Relax both problems by exploiting redundancy
Grasp representation as constraint envelopes
Grasp Synthesis
# 11 © Robert Krug et al. @RSS – Grasp Envelopes for Constraint-based Robot Motion Planning and Control (Jul 16, 2015)
2.
Exploiting Redundancy
Where (synthesis) and how (motion planning) to grasp the target object?
Central Idea: Relax both problems by exploiting redundancy
Grasp representation tailored to constrained optimal control
reactive motion planning- and generation
Motion Planning
# 12 © Robert Krug et al. @RSS – Grasp Envelopes for Constraint-based Robot Motion Planning and Control (Jul 16, 2015)
Exploiting Redundancy
Where (synthesis) and how (motion planning) to grasp the target object?
Pros:
• Fast -> no planning delays
• Reactive
• Incorporate sensory feedback
• Incorporate prior knowledge
• Leverage redundancy
Suggested solution: constraint-based grasp- and environment description
and locally optimal motion control/generation
Cons:
• Approximated environment
• Local (depending on control)
# 13 © Robert Krug et al. @RSS – Grasp Envelopes for Constraint-based Robot Motion Planning and Control (Jul 16, 2015)
2.
Exploiting Redundancy
OFFLINE ONLINE
Grasping Pipeline
define constraint
envelope templates perception
prune envelopes
motion generation
& control
robust grasp execution
# 14 © Robert Krug et al. @RSS – Grasp Envelopes for Constraint-based Robot Motion Planning and Control (Jul 16, 2015)
Robust Grasp Execution
3
# 15 © Robert Krug et al. @RSS – Grasp Envelopes for Constraint-based Robot Motion Planning and Control (Jul 16, 2015)
3. Robust Grasp Execution
How about the introduced approximation and/or uncertainty errors?
Compensate with low pose sensitivity gripper [Tincani et al., IROS, 2012]
# 16 © Robert Krug et al. @RSS – Grasp Envelopes for Constraint-based Robot Motion Planning and Control (Jul 16, 2015)
Preliminary Results
4
# 17 © Robert Krug et al. @RSS – Grasp Envelopes for Constraint-based Robot Motion Planning and Control (Jul 16, 2015)
4.
Preliminary Results
Local optimal control scheme
[Mansard, ICAR, 2009], [Kanoun, TRO, 2011]
# 18 © Robert Krug et al. @RSS – Grasp Envelopes for Constraint-based Robot Motion Planning and Control (Jul 16, 2015)
4. Preliminary Results
KUKA Innovation Award Finalist: Advanced Picking & Palletizing (APPLE)
# 19 © Robert Krug et al. @RSS – Grasp Envelopes for Constraint-based Robot Motion Planning and Control (Jul 16, 2015)
Discussion
5
# 20 © Robert Krug et al. @RSS – Grasp Envelopes for Constraint-based Robot Motion Planning and Control (Jul 16, 2015)
5.
Discussion
Take away: Achieve fast and reactive grasping/manipulation behaviors by
exploiting redundancy via constraint-based grasp representations tailored
to optimal control schemes
Open Issues:
• “More global” solution would be nice MPC control
• Incorporate sensory feedback
• Learn from demonstrations and/or experience
• …
Mobile Robotics and Olfaction Lab,
AASS, Örebro University
# 21 RSS 2015 July 16th 2015, Rome, Italy
Robert Krug Todor Stoyanov Achim J. Lilienthal [email protected] [email protected] [email protected]
# 22 © Robert Krug et al. @RSS – Grasp Envelopes for Constraint-based Robot Motion Planning and Control (Jul 16, 2015)
References
6
# 23 © Robert Krug et al. @RSS – Grasp Envelopes for Constraint-based Robot Motion Planning and Control (Jul 16, 2015)
6.
R. Krug, T. Stoyanov, M. Bonilla, V. Tincani, N. Vaskevicius, G. Fantoni, A. Birk, A. J. Lilienthal and A. Bicchi. Velvet Fingers: Grasp Planning and Execution for an Underactuated Gripper with Active Surfaces. Proc. of the IEEE Int. Conf. on Robotics and Automation, 2013, pp. 3669-3675.
T. Stoyanov, N. Vaskevicius, C. A. Mueller, T. Fromm, R. Krug, V. Tincani, R. Mojtahedzadeh, S. Kunaschk, R. Mortensen Ernits, D. R. Canelhas, M. Bonilla, S. Schwertfeger, M. Bonini, H. Halfar, K. Pathak, M. Rohde, G. Fantoni, A. Bicchi, A. Birk, A. J. Lilienthal and W. Echelmeyer. No More Heavy Lifting: Robotic Solutions to the Container Unloading Problem. IEEE Robotics & Automation Magazine, submitted, 2015.
D. Berenson, R. Diankov, K. Nishiwaki, S. Kagami and J. Kuffner Grasp Planning in Complex Scenes. Proc. of the IEEE/RAS Int. Conf. on Hummanoid Robots, 2007, pp. 42-48
References
# 24 © Robert Krug et al. @RSS – Grasp Envelopes for Constraint-based Robot Motion Planning and Control (Jul 16, 2015)
6.
V. Tincani, M. G. Catalano, E. Farnioli, M. Garabini, G. Grioli, G. Fantoni and A. Bicchi. Velvet Fingers: A Dexterous Gripper with Active Surfaces. Proc. of the IEEE/RAS Int. Conf. on Intelligent Robots and Systems, 2012, pp. 1257-1263.
N. Mansard, O. Stasse, P. Evrard and A. Kheddar. A Versatile Generalized Inverted Kinematics Implementation for Collaborative Working Humanoid Robots: The Stack of Tasks. Proc. of the Int. Conf. on Advanced Robotics, 2009, pp. 1-6.
O. Kanoun, F. Lamiraux and P.-B. Wieber Kinematic Control of Redundant Manipulators: Generalizing the Task-Priority Framework to Inequality Task. IEEE Transactions on Robotics, 2011, 27:4, pp. 785-792
References