Biorobotics Laboratory BioRob

Adding Vision To Snake Robot
Benoit Rat - Semester Project 2006-2007
Supervisor: Dr. Francois Fleuret, CVLAB
Collaborators: Prof. Auke Jan Ijspeert - Alessandro Crespi.

Abstract

The Adding Vision to Snake Robot project is focused on designing algorithms for stimulus tracking and obstacle avoidance for the robot moving in cluttered environments. The algorithms needed to be tested on the real robot in engineered environments of increasing complexity.

This project focus on primary level of artificial vision such as color segmentation. A robust algorithm for color based tracking supporting changes of illumination is implemented. By using this algorithm the robot is able to follow a pink ball in real-time generating specific stimuli.

A more complex artificial vision process has also been developed during the resarch phase. This second algorithm finds the geometrical transformation between two consecutive frames and thus deduces the relative position of our robot.

Documents

Report of the project (PDF, 6.4 MB)
Slides of the presentation (PDF, 570 KB)

Videos of Color Tracking

Offline (627 KB)	Inside view (3.9 MB)
An offline video of the color tracking algorithm. The four screens represent respectively: the input frame, the probability map, the mask, the output result.	Online color tracking with view from the robot. This video does not correspond to the "top view" videos but is really similar
Top View 1 (6.1 MB)	Top View 2 (843KB)

Screenshots of Matching Patch Algorithm

Two different screenshots of the Matching Patch Algorithm: (For more informations look at the presentation or the report)

Links

This project was supervised by François Fleuret from the CVLAB (Computer Vision Lab)
The source code was developped using: OpenCV Library