Finding Visual Landmarks with E-Puck robots Loc SEGAPELLI - - PowerPoint PPT Presentation

finding visual landmarks with e puck robots
SMART_READER_LITE
LIVE PREVIEW

Finding Visual Landmarks with E-Puck robots Loc SEGAPELLI - - PowerPoint PPT Presentation

Finding Visual Landmarks with E-Puck robots Loc SEGAPELLI Introduction Webots A group of e-puck robots Each one has a camera They are looking for a blue light They must use their camera only! Collaboration Example


slide-1
SLIDE 1

Finding Visual Landmarks with E-Puck robots

Loïc SEGAPELLI

slide-2
SLIDE 2

Introduction

Webots A group of e-puck robots Each one has a camera They are looking for a blue light They must use their camera only! Collaboration

slide-3
SLIDE 3

Example

slide-4
SLIDE 4

Algorithm

Robot turn 360º , takes 6 photos One robot sees the blue light:

Turns on its red leds Waits Goes to the light source

slide-5
SLIDE 5

Algorithm

Another robot sees red leds:

Turns on its red leds -> chain reaction Waits Goes to the red light source

slide-6
SLIDE 6

Blue light detection 1

slide-7
SLIDE 7

Blue light detection 2

slide-8
SLIDE 8

Blue light detection 3

slide-9
SLIDE 9

Blue light detection 4

slide-10
SLIDE 10

Blue light detection 5

slide-11
SLIDE 11

Blue light detection 6

slide-12
SLIDE 12

Blue light detection 7

slide-13
SLIDE 13

Image segmentation

One picture = 1 min! Get the most from one photo Vertical image segmentation

slide-14
SLIDE 14

Image segmentation

Position correction: robots turn where there is the most blue / red pixels

slide-15
SLIDE 15

Image segmentation 1

slide-16
SLIDE 16

Image segmentation 2

slide-17
SLIDE 17

Image segmentation 3

slide-18
SLIDE 18

Image segmentation 4

slide-19
SLIDE 19

The whole state machine…

Exploration Detection

slide-20
SLIDE 20

Real camera pictures

30x30 pixels, should not be a handicap

slide-21
SLIDE 21

Conclusion

  • No real-robot experimentation

+ Interesting image segmentation strategy + Simple state machine