1. NUS CS 5247 David Hsu Minkowski Sum Gokul Varadhan

2. 2 Last Lecture  Configuration space workspace configuration space

3. 3 Problem – Configuration Space of a Translating Robot  Input: Polygonal moving object translating in 2-D workspace Polygonal obstacles  Output: configuration space obstacles represented as polygons

4. 4 Configuration Space of a Translating Robot Workspace Configuration Space x y Robot Obstacle C- obstacle Robot  C-obstacle is a polygon.

5. 5 Minkowski Sum A B

6. 6

7. 7

8. 8

9. 9 Configuration Space Obstacle O - R Obstacle O Robot R C- obstacle C-obstacle is O - R Classic result by Lozano-Perez and Wesley 1979

10. 10 Properties of Minkowski Sum  Minkowski sum of boundary of P and boundary of Q is a subset of boundary of  Minkowski of two convex sets is convex PQPQ

11. 11 Minkowski sum of convex polygons  The Minkowski sum of two convex polygons P and Q of m and n vertices respectively is a convex polygon P + Q of m + n vertices. The vertices of P + Q are the “sums” of vertices of P and Q. =

12. 12 Gauss Map  Gauss map of a convex polygon Edge  point on the circle defined by the normal Vertex  arc defined by its adjacent edges

13. 13 Gauss Map Property of Minkowski Sum  p+q belongs to the boundary of Minkowski sum only if the Gauss map of p and q overlap.

14. 14 Computational efficiency  Running time O(n+m)  Space O(n+m)

15. 15 Minkowski Sum of Non-convex Polygons  Decompose into convex polygons (e.g., triangles or trapezoids),  Compute the Minkowski sums, and  Take the union  Complexity of Minkowski sum O(n 2 m 2 )

16. 16 Worst case example  O(n 2 m 2 ) complexity 2D example Agarwal et al. 02

17. 17 3D Minkowski Sum  Convex case O(nm) complexity Many methods known for computing Minkowski sum in this case  Convex hull method Compute sums of all pairs of vertices of P and Q Compute their convex hull O(mn log(mn)) complexity  More efficient methods are known [Guibas and Seidel 1987]

18. 18 3D Minkowski Sum  Non-convex case O(n 3 m 3 ) complexity  Computationally challenging  Common approach resorts to convex decomposition

19. 19 3D Minkowski Sum Computation  Two objects P and Q with m and n convex pieces respectively Compute mn pairwise Minkowski sums between all pairs of convex pieces Compute the union of the pairwise Minkowski sums  Main bottleneck Union computation mn is typically large (tens of thousands) Union of mn pairwise Minkowski sums has a complexity close to O( m 3 n 3 )  No practical algorithms known for exact Minkowski sum computation

20. 20 Minkowski Sum Approximation  We developed an accurate and efficient approximate algorithm [Varadhan and Manocha 2004]  Provides certain “geometric” and “topological” guarantees on the approximation Approximation is “close” to the boundary of the Minkowski sum It has the same number of connected components and genus as the exact Minkowski sum

21. 21 Brake Hub (4,736 tris) Union of 1,777 primitives Rod (24 tris)

22. 22 Union of 4,446 primitives Anvil (144 tris) Spoon (336 tris)

23. 23 Knife (516 tris) Scissors (636 tris) Union of 63,790 primitives

24. 24 444 tris1,134 tris

25. 25 Union of 66,667 primitives

26. 26 Cup (1,000 tris) Cup Offset Gear 2,382 tris) Gear Offset Offsetting

27. 27 Configuration Space Approximation - 3D Translation Obstacle O Robot R O - R C-obstacle

28. 28 Assembly Obstacle Robot

29. 29 Assembly Roadmap 16 secs Path Search 0.22 secs Start Goal Obstacle

30. 30 Assembly

31. 31 Path in Configuration Space Start Goal Path

32. 32 Other Applications  Minkowski sums and configuration spaces have also been used for Morphing Tolerance Analysis Knee/Joint Modeling Interference Detection Penetration Depth Packing

33. 33 Applications - Dynamic Simulation Kim et al. 2002 Interference Detection Penetration Depth Computation

34. 34 Morphing A B Morph

35. 35 Applications - Packing

36. 36 Next lecture  Configuration space of a polygonal robot capable of translation and rotation