Zone Using OpenCV Erosion and Image Contractor And Platooning Approach to form a chain

Transcription

1 Bay Modeling of Biscay the Secure Project Zone Using OpenCV Erosion and Image Contractor And Platooning Approach to form a chain MEA 09/11/2017 Khadimoullah Vencatasamy -- Luc Jaulin Alexandre Chapoutot 1

2 Summary Ellipse Strategy Erosion OpenCV Image contractor Platonning Khadimoullah Vencatasamy -- Luc Jaulin 2

3 Presentation Khadimoullah Vencatasamy -- Luc Jaulin 3

4 Presentation Two problems: Find a strategy to cover all the area Ellipsoid strategy Define a criteria to validate the strategy algorithm Interval Analysis Khadimoullah Vencatasamy -- Luc Jaulin 4

5 Formalism of the problem Khadimoullah Vencatasamy -- Luc Jaulin 5

6 Ellipsoid Strategy For the i-th robot, Khadimoullah Vencatasamy -- Luc Jaulin 6

7 Ellipsoid Strategy Khadimoullah Vencatasamy -- Luc Jaulin 7

8 Trail Khadimoullah Vencatasamy -- Luc Jaulin 8

9 Erosion Khadimoullah Vencatasamy -- Luc Jaulin 9

10 Erosion Rectangle Kernel Khadimoullah Vencatasamy -- Luc Jaulin 10

11 Erosion Cross Kernel Khadimoullah Vencatasamy -- Luc Jaulin 11

12 Erosion Ellipse Kernel Khadimoullah Vencatasamy -- Luc Jaulin 12

13 Trail with Circles Khadimoullah Vencatasamy -- Luc Jaulin 13

14 Erosion vs Trail With Circles Khadimoullah Vencatasamy -- Luc Jaulin 14

15 Erosion vs Trail With Circles Khadimoullah Vencatasamy -- Luc Jaulin 15

16 Image Contractor Binary Image Compute the Integral Image SIVIA Subpavi ng IN = White Out = Black A C S = D - B - C + A B Test(S) { If S = 0: D Then OUT If S = NbPixelInRect : Then IN Else: UNKNOWN } Khadimoullah Vencatasamy -- Luc Jaulin 16

17 Image Contractor Khadimoullah Vencatasamy -- Luc Jaulin 17

18 Image Contractor Khadimoullah Vencatasamy -- Luc Jaulin 18

19 Image Contractor Khadimoullah Vencatasamy -- Luc Jaulin 19

20 Final Results Khadimoullah Vencatasamy -- Luc Jaulin 20

21 Next Work Use dilate function in pyibex instead of OpenCV Regulation of the robots Non causal case Intruder Interception Simulation / Experiment Khadimoullah Vencatasamy -- Luc Jaulin 21

22 Next Work Use dilate function in pyibex instead of OpenCV Regulation of the robots Non causal case Intruder Interception Simulation / Experiment Khadimoullah Vencatasamy -- Luc Jaulin 22

23 Formation of a chain between two points A B A B Khadimoullah Vencatasamy -- Luc Jaulin 23

24 Platooning Leader Follower Follower Follower The leader decides the path and communicate its position, speed, the path it has been chosen to all the followers. The follower got the data of the leader and the distance and speed of its predecessor. Khadimoullah Vencatasamy -- Luc Jaulin 24

25 Platooning There is the longitudinale control There is the lateral Control D0 D1 DL i Leader Follower Follower DL i+1 Khadimoullah Vencatasamy -- Luc Jaulin 25

26 Platooning Differents approaches exists which can be divided in two categories: Local Approach The robot does not rely on the other robots which increase the robustness of the system The cost to regulate the robot is low The stablitity is reduced due to the time to converge to the target position and the accumulation of the errors. Global Approach Need strong communication between each robots Cost to regulate the system is high Highly stable Robustness is reduced, if one element of the platoon fails, the chain will break. Khadimoullah Vencatasamy -- Luc Jaulin 26

27 Platooning We consider in our case that you could have a large number of robot and in our environment, communication could be a problem. We should adopt a local architecture Stability will be the main problem of the system Define a command law to avoid the oscillations Lyapunov function Khadimoullah Vencatasamy -- Luc Jaulin 27

28 Platooning Two Scenarii The robot maintain a distance considering its predecessor with a visual regulation and knowing the path it should follow The robots are following their path and if they see and communicate rarely with the opposite coming robot Khadimoullah Vencatasamy -- Luc Jaulin 28