1. Robosoccer Team MI20 presents … Supervisors Albert Schoute Mannes Poel Current team members Paul de Groot Roelof Hiddema Mobile Intelligence Twente

2. Robot soccer as a scientific “playing field” Interdisciplinary Hardware & Software Sensing & Control Image processing Motion planning Multi-agent collaboration Communication Artificial intelligence International Championships (FIRA, Robocup) Congresses

3. Mission Impossible ?

4. International leagues Robocup Humanoid Small size Middle size Four-Legged Rescue Junior Simulation @Home FIRA HuroSot KheperaSot MiroSot  UT team NaroSot QuadroSot RoboSot SimuroSot

5. Robocup Humanoid 2 vs. 2 (Osaka 2005)

6. FIRA Humanoid (Vienna 2003)

7. Robocup Middle League

8. Robocup Small League

9. FIRA Mirosot (11 vs. 11)

10. Games between teams of 5, 7 or 11 robots Camera’s above the field observe the playing Computers control the robots wirelessly FIRA Mirosot competitie

11. MiroSot robots Maximal dimensions: 7.5 x 7.5 x 7.5 cm Two-wheeled differential drive robots Board-computer controls wheel velocities

12. Impression of EC 2005 @ UT

13. Twente’s robosoccer team Started in 2002: Missing Impossible Mission Impossible Mobile Intelligence

14. Generations of students 1 st team Ljubljana 2003 4 th team Vienna 2006

15. Generations of robots

16. Home base

17. Computer control

18. Localization Robots have color patches on top Design is free, except for obligatory team color Design choice: identical or different patterns per robot? Identical  makes recognition simpler, but robots must be tracked

19. Vision

20. Camera image

21. Color segmentation

22. Color separation

23. Region clustering

24. Camera calibration Lens distortion

25. Image correction Remap feature points only

26. Correction of projective mapping Automatic field calibration by 4 known markers

27. Correct for parallax

28. Tracking

29. State estimation ) θ) θ x y (x,y)

30. Result on the screen

31. Motion Control Robots have local PID velocity controllers Motion commands  wheel speeds (v r, v l ) cq. lin. & ang. velocities (v,  ) Kinematic robot model Higher speeds: account for dynamics!

32. Motion Planning Driving fast to play the ball while avoiding obstacles …

33. Strategy ? The team’s magic

34. System design ? The team’s pain

35. (Re)designing for the winning team Initial MI20 multi-agent system architecture:

36. 1 st team motion controller Solve the parking problem: move to “pose” (x, y,  )

37. … while avoiding obstacles Vector Field Histogram Corresponding Histogram Local method:

38. Trying out in de simulator

39. Shoot and score!

40. Shoot and miss!

41. Improvements RealPrediction 1 22 1 1 2 2 1 Avoid tracking errors by collision analysis

42. Collision prediction last 1 pred 1 pred 2 γ last 2

43. Collision state correction corr 1 corr 2 last 2 pred 2 last 1 pred 1

44. Collision response model B VBVB P n A VAVA

45. Collision response (cont.) B VBVB ωBωB A VAVA ωAωA P

46. Improving strategy Choosing optimal offensive / defensive positions

47. Improved system structure Complete software revision Reduced thread concurrency Simplified interprocess communication Current O.S. Linux Fedora Core 4

48. Coming soon …

49. TU Vienna Parade

50. Questions?