distributed localization and control of a group of
play

Distributed localization and control of a group of underwater robots - PDF document

Distributed localization and control of a group of underwater robots using contractor programming L. Jaulin, S. Rohou, J. Nicola, M. Saad, F. Le Bars and B. Zerr. SWIM15 Prague, June 9-11, 2015 ENSTA Bretagne, OSM, LabSTICC. Video of the


  1. Distributed localization and control of a group of underwater robots using contractor programming L. Jaulin, S. Rohou, J. Nicola, M. Saad, F. Le Bars and B. Zerr.

  2. SWIM’15 Prague, June 9-11, 2015 ENSTA Bretagne, OSM, LabSTICC. Video of the presentation https://youtu.be/q6F7WDCcf2A

  3. 1 Scout project

  4. Goal : (i) coordination of underwater robots ; (ii) collabo- rative behavior.

  5. Supervisors : L. Jaulin, C. Aubry, S. Rohou, B. Zerr, J. Nicola, F. Le bars Compagny : RTsys (P. Raude) Students : G. Ricciardelli, L. Devigne, C. Guillemot, S. Pommier, T. Viravau, T. Le Mezo, B. Sultan, B. Moura, M. Fadlane, A. Bellaiche, T. Blanchard, U. Da rocha, G. Pinto, K. Machado.

  6. 2 Controller

  7. 3 Localization problem

  8. Range only Based on interval analysis Robust with respect to outliers Distributed computation Low rate communication We propose here to use a contractor programming ap- proach

  9. 4 Matrices and contractors

  10. linear application matrices → � � � α = 2 a + 3 h 2 3 L : → A = γ = h − 5 a 1 − 5 We have a matrix algebra and Matlab. We have: var ( L ) = { a, h } , covar ( L ) = { α, γ } . But we cannot write: var ( A ) = { a, h } , covar ( A ) = { α, γ } .

  11. constraint → contractor a · b = z →

  12. Contractor fusion � a · b = z → C 1 b + c = d → C 2 Since b occurs in both constraints, we fuse the two con- tractors as: C = C 1 × C 2 ⌋ (2 , 1) = C 1 |C 2 (for short)

  13. 5 Localization with contractors

  14. x j h +1 = f ( x j h , u j h ) , h ∈ { k − ¯ h, . . . , k } = � The observer: C k,j h,...,k } C j x h ∈{ k − ¯ x ( h ) � � var ( C k,j x ) = var ( C k,j x j h , . . . , x j k , x j x ( h ) ) = . k − ¯ k +1

  15. z j h = h ( x j h ) � � x ∩ � { q 1 } Observer RSO: C k,j x ,z = C k,j C k,j x | C h,j . z h ∈{ k − ¯ h,...,k }

  16. To get an outer approximation of set ( C k,j x ,z ) , we need a paver. We can also obtain an inner approximation using separators [SMART 2015].

  17. t = 0

  18. t = 0 . 1

  19. t = 0 . 2

  20. t = 0 . 3

  21. t = 0 . 4

  22. t = 0 . 5

  23. t = 0 . 6

  24. t = 0 . 7

  25. t = 0 . 8

  26. t = 0 . 9

  27. t = 1 . 0

  28. t = 1 . 1

  29. t = 1 . 2

  30. t = 1 . 3

  31. t = 1 . 4

  32. 6 Distributed localization

  33. y j,ℓ h = g ( x j h , x ℓ h )

  34. Observer: � � � { q 1 } C k,j = C k,j C k,j x | C h,j ∩ x ,z x z h ∈{ k − ¯ h,...,k } � � � { q 2 } C k,j x | C h,j,ℓ ∩ y h ∈{ k − ¯ h,...,k }

  35. 7 Singularity

  36. 8 Test case

  37. QT/C++ code available at http://www.ensta-bretagne.fr/jaulin/easibex.html

  38. 9 Tests

Recommend


More recommend