Bimanual Regrasping Balaguer et al. Bimanual Regrasping Introduction from Unimanual Machine Learning Algorithm Overview Vision Grasp Synthesis Optimization Benjamin Balaguer and Stefano Carpin Experiments Comparison University of California, Merced Efficiency Examples Conclusion IEEE International Conference on Robotics and Automation Grasping and Manipulation Session (WeD02) May 16, 2012 1 / 12
Regrasping An Overview Bimanual Regrasping Regrasping Problem Balaguer et al. Modify an object’s configuration when limits/constraints are reached Introduction Algorithm Regrasping in the robotics literature Overview Vision In-hand Regrasping Grasp Synthesis Optimization Depends on dexterous end-effector Experiments On-surface Regrasping Comparison Efficiency Slow and non-human-like Examples Conclusion Proposed solution: bimanual regrasping Requires dual manipulators Human inspired Focus on speed and efficiency 2 / 12
Regrasping An Overview Bimanual Regrasping Regrasping Problem Balaguer et al. Modify an object’s configuration when limits/constraints are reached Introduction Algorithm Regrasping in the robotics literature Overview Vision In-hand Regrasping Grasp Synthesis Optimization Depends on dexterous end-effector Experiments On-surface Regrasping Comparison Efficiency Slow and non-human-like Examples Conclusion Proposed solution: bimanual regrasping Requires dual manipulators Human inspired Focus on speed and efficiency 2 / 12
Regrasping An Overview Bimanual Regrasping Regrasping Problem Balaguer et al. Modify an object’s configuration when limits/constraints are reached Introduction Algorithm Regrasping in the robotics literature Overview Vision In-hand Regrasping Grasp Synthesis Optimization Depends on dexterous end-effector Experiments On-surface Regrasping Comparison Efficiency Slow and non-human-like Examples Conclusion Proposed solution: bimanual regrasping Requires dual manipulators Human inspired Focus on speed and efficiency 2 / 12
Algorithm Overview Bimanual Problem Definition: Regrasping Balaguer et al. Given an object only reachable by the right manipulator, move it to an area only reachable by the left manipulator Introduction Algorithm Overview Vision Grasp Synthesis Optimization Experiments Comparison Efficiency Examples Conclusion 3 / 12
Algorithm Overview Bimanual Problem Definition: Regrasping Balaguer et al. Given an object only reachable by the right manipulator, move it to an area only reachable by the left manipulator Introduction Algorithm Problem Solution: Overview Vision Minimize execution time Grasp Synthesis Optimization Cast as an optimization problem Experiments Comparison Efficiency Examples Conclusion 3 / 12
Algorithm Overview Bimanual Problem Definition: Regrasping Balaguer et al. Given an object only reachable by the right manipulator, move it to an area only reachable by the left manipulator Introduction Algorithm Problem Solution: Overview Vision Minimize execution time Grasp Synthesis Optimization Cast as an optimization problem Experiments Comparison Efficiency Examples Conclusion 3 / 12
Algorithm Overview Bimanual Problem Definition: Regrasping Balaguer et al. Given an object only reachable by the right manipulator, move it to an area only reachable by the left manipulator Introduction Algorithm Problem Solution: Overview Vision Minimize execution time Grasp Synthesis Optimization Cast as an optimization problem Experiments Comparison Efficiency Examples Conclusion 3 / 12
Algorithm Overview Bimanual Problem Definition: Regrasping Balaguer et al. Given an object only reachable by the right manipulator, move it to an area only reachable by the left manipulator Introduction Algorithm Problem Solution: Overview Vision Minimize execution time Grasp Synthesis Optimization Cast as an optimization problem Experiments Comparison Efficiency Examples Conclusion 3 / 12
Algorithm Overview Bimanual Problem Definition: Regrasping Balaguer et al. Given an object only reachable by the right manipulator, move it to an area only reachable by the left manipulator Introduction Algorithm Problem Solution: Overview Vision Minimize execution time Grasp Synthesis Optimization Cast as an optimization problem Experiments Comparison Efficiency Examples Conclusion 3 / 12
Reusing Unimanual Grasper Image Processing [Saxena et al. 2008] Bimanual Regrasping Balaguer et al. Introduction Algorithm Overview Vision Grasp Synthesis Optimization Start from object-extracted image-space Experiments Apply Canny edge detector Comparison Efficiency Examples Apply [Saxena et al. 2008], keeping ∀ p i , P ( z i = 1) > 0 . 9 Conclusion Use heuristic to choose two good grasping points 4 / 12
Reusing Unimanual Grasper Image Processing [Saxena et al. 2008] Bimanual Regrasping Balaguer et al. Introduction Algorithm Overview Vision Grasp Synthesis Optimization Start from object-extracted image-space Experiments Apply Canny edge detector Comparison Efficiency Examples Apply [Saxena et al. 2008], keeping ∀ p i , P ( z i = 1) > 0 . 9 Conclusion Use heuristic to choose two good grasping points 4 / 12
Reusing Unimanual Grasper Image Processing [Saxena et al. 2008] Bimanual Regrasping Balaguer et al. Introduction Algorithm Overview Vision Grasp Synthesis Optimization Start from object-extracted image-space Experiments Apply Canny edge detector Comparison Efficiency Examples Apply [Saxena et al. 2008], keeping ∀ p i , P ( z i = 1) > 0 . 9 Conclusion Use heuristic to choose two good grasping points 4 / 12
Reusing Unimanual Grasper Image Processing [Saxena et al. 2008] Bimanual Regrasping Balaguer et al. Introduction Algorithm Overview Vision Grasp Synthesis Optimization Start from object-extracted image-space Experiments Apply Canny edge detector Comparison Efficiency Examples Apply [Saxena et al. 2008], keeping ∀ p i , P ( z i = 1) > 0 . 9 Conclusion Use heuristic to choose two good grasping points � | P ( z i =1)+ P ( z j =1) | � arg max i , j � p i − p j � ∀ i , j ∈ R 2 Constraint: P G i ( z ), P G j ( z ) ≥ 5cm 4 / 12
Reusing Unimanual Grasper Grasp Synthesis [Balaguer et al. 2010] Bimanual Regrasping Image Processing Balaguer et al. Left Point Introduction Point Cloud Image Right Point Algorithm Overview Vision Grasp Synthesis Optimization Experiments Comparison Efficiency Orientation Nearest Examples Classification Neighbor Search Estimation Conclusion Training Data 5 / 12
Moving the Object to a Regrasping Configuration Mathematical Derivation Bimanual Regrasping Balaguer et al. Introduction Algorithm Overview Vision Grasp Synthesis Optimization Experiments Comparison Efficiency Examples Assume regrasping is given by P G Ropt and R G Conclusion Ropt V G = ( R G Ropt )[( R G Rini ) T ( P G Lini − P G Rini )] P G Lopt = P G Ropt + V G R G Lopt = ( R G Ropt )[( R G Rini ) T ( R G Lini )] 6 / 12
Moving the Object to a Regrasping Configuration Mathematical Derivation Bimanual Regrasping Balaguer et al. Introduction Algorithm Overview Vision Grasp Synthesis Optimization Experiments Comparison Efficiency Examples Assume regrasping is given by P G Ropt and R G Conclusion Ropt V G = ( R G Ropt )[( R G Rini ) T ( P G Lini − P G Rini )] P G Lopt = P G Ropt + V G R G Lopt = ( R G Ropt )[( R G Rini ) T ( R G Lini )] 6 / 12
Moving the Object to a Regrasping Configuration Mathematical Derivation Bimanual Regrasping Balaguer et al. Introduction Algorithm Overview Vision Grasp Synthesis Optimization Experiments Comparison Efficiency Examples Assume regrasping is given by P G Ropt and R G Conclusion Ropt V G = ( R G Ropt )[( R G Rini ) T ( P G Lini − P G Rini )] P G Lopt = P G Ropt + V G R G Lopt = ( R G Ropt )[( R G Rini ) T ( R G Lini )] 6 / 12
Moving the Object to a Regrasping Configuration Mathematical Derivation Bimanual Regrasping Balaguer et al. Introduction Algorithm Overview Vision Grasp Synthesis Optimization Experiments Comparison Efficiency Examples Assume regrasping is given by P G Ropt and R G Conclusion Ropt V G = ( R G Ropt )[( R G Rini ) T ( P G Lini − P G Rini )] P G Lopt = P G Ropt + V G R G Lopt = ( R G Ropt )[( R G Rini ) T ( R G Lini )] 6 / 12
Moving the Object to a Regrasping Configuration Mathematical Derivation Bimanual Regrasping Balaguer et al. Introduction Algorithm Overview Vision Grasp Synthesis Optimization Experiments Comparison Efficiency Examples Assume regrasping is given by P G Ropt and R G Conclusion Ropt V G = ( R G Ropt )[( R G Rini ) T ( P G Lini − P G Rini )] P G Lopt = P G Ropt + V G R G Lopt = ( R G Ropt )[( R G Rini ) T ( R G Lini )] 6 / 12
Moving the Object to a Regrasping Configuration Mathematical Derivation Bimanual Regrasping Balaguer et al. Introduction Algorithm Overview Vision Grasp Synthesis Optimization Experiments Comparison Efficiency Examples Assume regrasping is given by P G Ropt and R G Conclusion Ropt V G = ( R G Ropt )[( R G Rini ) T ( P G Lini − P G Rini )] P G Lopt = P G Ropt + V G R G Lopt = ( R G Ropt )[( R G Rini ) T ( R G Lini )] 6 / 12
More recommend
