1 Computing Deform Closure Grasps K. “Gopal” Gopalakrishnan, Ken Goldberg IEOR and EECS, UC Berkeley
2 Related Work D-Space and Deform Closure Two-point deform closure grasps Algorithm Outline
3 Workholding: Rigid parts Contact Mechanics: Number of contacts –[Reuleaux, 1876], [Somoff, 1900] –[Mishra, Schwarz, Sharir, 1987], –[Nguyen, 1988] –[Markenscoff, Papadimitriou, 1990] –[Han, Trinkle, Li, 1999] Immobility, 2 nd Order Form Closure –[Rimon, Burdick, 1995, 1998] –[Ponce, Burdick, Rimon, 1995] [Mason, 2001]
4 Workholding: Rigid parts Caging Grasps –[Rimon, Blake, 1999] Summaries of results –[Bicchi, Kumar, 2000] –[Mason, 2001] C-Spaces for closed chains –[Milgram, Trinkle, 2002] Fixturing hinged parts –[van der Stappen et al, 2002] Antipodal Points for Curved Parts –[Jia 2002]
5 C-Space C-Space (Configuration Space): [Lozano-Perez, 1983] Dual representation Each degree of part freedom is one C-space dimension. y x /3 (5,4) y x (5,4,- /3) Physical space C-Space
6 Elastic Fingers, Soft Contacts [Hanafusa and Asada, 1982] [Salisbury and Mason, 1985] [Kim, Hirai, Inoue, 2003] Physical Models [Joukhadar, Bard, Laugier, 1994] Bounded Force Closure [Wakamatsu, Hirai, Iwata, 1996] Learned Grasps of Deformable Objects [Howard, Bekey, 1999] Deformable Parts
7 Path Planning for flexible wires, Elastic Sheets and Bodies [Kosuge et al, 1995] [Kavraki et al, 1998, 2000] [Amato et al, 2001] [Moll and Kavraki, 2004] Robot manipulation [Henrich and Worn, 2000] Robust manipulation with Vision [Hirai, Tsuboi, Wada 2001]
8 Modified spring-mass modeling [Yasuda, Yokoi, 2001] Holding sheet metal parts with dexterous parameter estimation [Ceglarek et al, 2002] Haptic rendering of deformations [Laycock, Day, 2003] Simulation of human hand deformation [Latombe et al, 2003] Deformable Parts
9 Related Work D-Space and Deform Closure Two-point deform closure grasps Algorithm Outline
10 Inspiration: Earlier work Grasping at concavities: –2D v-grips –3D v-grips
11 Inspiration: Earlier work Unilateral fixtures –Sheet metal parts.
12 Deformable parts “Form closure” insufficient: The part can always escape if deformed.
13 Inspiration: Earlier work Surgery simulation.
14 Lack of definition of fixtures/grasps for deformable parts. Generalization of C-Space. Inspiration
15 C-Space: Avoiding Collisions Finger body A in physical space is mapped on to the C- obstacle, CA. In C-space, the part shrinks to a point and the obstacle grows accordlingly CA c = C free. Physical space C-Space x y
16 Form Closure A part is grasped in Form Closure if any infinitesimal motion results in collision. Form Closure is equivalent to an isolated point in C-free. Physical space C-Space x y
17 Part Model Planar Polygonal Boundary Triangular Mesh Nodes (like hinges) Edges (like struts) Elements (deformable)
18 Deformation Space (D-Space) Each node has 2 DOF Analogous to configurations in C-Space D-Space: 2N-dimensional space of mesh node positions. point q in D-Space is a “deformation” q 0 is initial (undeformed) point 30-dimensional D-space
19 D-Space: Example Example part: 3-node mesh, 2 fixed. D-Space: 2D x y Physical space D-Space q0q0
20 Topology violating deformation Undeformed part Allowed deformation Self Collisions
21 D T : Topology Preserving Subspace x y Physical space D-Space D T D-Space. DTDT DTC:DTC:
22 D-Obstacles x y Physical space D-Space Collision of any mesh element with obstacle. A Physical obstacle A i defines a deformation- obstacle DA i in D-Space. A1A1 DA 1
23 D-Space: Example Physical space x y D-Space Like C free, we define D free. D free = D T [ (DA i C )]
24 With more nodes: D free Slice with only node 5 moving. Part and mesh x y Slice with only node 3 moving. x 3 y 3 x 5 y 5 x 5 y 5 x 5 y 5 Physical space D-Space D free = D T [ (DA i C )]
25 Displacement between a pair of deformations q 0 : part’s nominal shape. X = q - q 0 : vector of nodal translations. Equivalent to moving origin in D-Space to q 0 and aligning nodal reference frames. q0q0 q X
26 Potential Energy Assume Linear Elasticity, Zero Friction K = FEM stiffness matrix. (2N 2N matrix for N nodes.) Forces at nodes: F = K X. Potential Energy: U(q) = (1/2) X T K X
27 Potential Energy “Surface” U : D free R 0 Equilibrium: q where U is at a local minimum. In absence of friction or inertia, part will come to rest at an equilibrium. Stable Equilibrium: q where U is at a strict local minimum. = “Deform Closure Grasp” q U(q)
28 Potential Energy Needed to Escape from a Stable Equilibrium Consider: Stable equilibrium q A, Equilibrium q B. “Capture Region”: K(q A ) D free, such that any configuration in K(q A ) returns to q A. Saddlepoints [Rimon, Blake, 1995] q A qBqB q U(q) K( q A )
29 U A (q A ) = Increase in Potential Energy needed to escape from q A. = minimum external work needed to escape from q A. U A : Quality Measure q A qBqB q U(q) UAUA Potential Energy Needed to Escape from a Stable Equilibrium K( q A )
30 U A : Example U A = 4 JoulesU A = 547 Joules
31 Deform-closure: Frame invariant. Equivalence of form-closure and deform-closure. Symmetries in D T and DA i to simplify computation. D-Space: Analysis and Results x y x y D1:D1: D2:D2:
32 Related Work D-Space and Deform Closure Two-point deform closure grasps Algorithm Outline
33 Given: Pair of contact nodes. Determine: Optimal jaw separation. Optimal? Problem Description M E n0n0 n1n1
34 Naïve Quality metric If Quality metric Q = U A. Maximum U A trivially at = 0
35 Quality metric Plastic deformation. Occurs when strain exceeds e L.
36 Plastic Deformation Stress Strain Plastic Deformation eLeL A B C
37 New Quality Metric Additional work U L done by jaws for plastic deformation. New Q = min { U A, U L }; maximize Q Stress Strain Plastic Deformation eLeL A B C D Grasp U L = U (D) – U (Grasp)
38 Additional input: e L : Elastic limit strain. : allowed error in quality metric. Additional assumptions: Sufficiently dense mesh !! Linear Elasticity. No Friction. Problem Description M, K E n0n0 n1n1
39 No collisions of mesh. Monotonicity of U between mesh nodes. q1q1 q2q2 q1q1 q2q2 U Assumptions
40 Related Work D-Space and Deform Closure Two-point deform closure grasps Algorithm Outline
41 Paths and Points of Interest in D-Space
42 q1q1 q2q2 U ( q 1 ) U ( q 2 ) Paths and Points of Interest in D-Space
43 Paths and Points of Interest in D-Space
44 Potential Energy vs. nini njnj k ij Potential Energy (U) Distance between FEM nodes Undeformed distance Expanding Contracting
45 Points of interest: contact at mesh nodes. Construct a graph: Each graph vertex = 1 pair of perimeter mesh nodes. p perimeter mesh nodes. O(p 2 ) graph vertices. Contact Graph
46 A B C E F G D Contact Graph: Edges Adjacent mesh nodes: A B C D E F G H H
47 Contact Graph Potential Energy Jaw 1 position Jaw 2 position
48 Contact Graph: Edges Non-adjacent mesh nodes:
49 Contact Graph Potential Energy Jaw 1 position Jaw 2 position Sparse graph.
50 Traversal with minimum increase in energy. FEM solution with two mesh nodes fixed. nini njnj Deformation at Points of Interest
51 U ( v(n i, n j ), ) Peak Potential Energy Given release path
52 Peak Potential Energy: All release paths U ( v *, )
53 U ( v o, ), U ( v*, ) Threshold Potential Energy U ( v*, ) U ( v o, ) U A ( ) U A ( ) = U ( v*, ) - U ( v o, )
54 U A ( ), U L ( ) Quality Metric U A ( ) U L ( ) Q ( )
55 Possibly exponential number of pieces. Sample in intervals of . Error bound on max. Q = * max { 0 (n i, n j ) * k ij } Numerical Sampling Q ( )
56 Calculate U L. To determine U A : Algorithm inspired by Dijkstra’s algorithm for sparse graphs. Fixed i
57 Algorithm for U A ( i )
58 Algorithm for U A ( i )
59 U Vertex v (traversed on path of minimum work) U(v) U(v*) Estimates for maximum U
60 Numerical Example Foam rubber part. Mesh and FEM using I-DEAS and ANSYS.
61 Numerical Example High elastic limit.
62 Numerical Example Undeformed = 10 mm. Optimal = 5.6 mm.
63 Future work Non-linear FEM. Mesh density: - Minimal meshes. - Special class of shapes? - Equivalent deformations between meshes?
64 Future work Mesh density: - Equivalent deformations between meshes? - Mapping D-spaces between meshes.
65 Future work Mesh density: - Error bounds. A A B B
66 Future work Optimal node selection.
67 Future work Friction