The work done in this master thesis concerns laser guided vehicles, and more specifically the autonomous navigation by following the edge of a road. The idea is that using two synchronized lasers on top of each other would increase the robustness of the road detection, compared to only one laser.
The algorithm was developed in Matlab, using a sequence of scans recorded on a snowy road with snow piles on the sides. The road detection uses the Hough transform for each scan. The relative position of the robot on the road is then estimated from the road edges detections. The control law is based on the “dog-rabbit principle”.
Simulations were done in Matlab, and a few tests have also been done on the MICA wheelchair. The algorithm was successful when following a straight corridor, but some more work has to be put into dealing with intersections, and on predicting the detection.
Source: Luleå University of Technology
Author: Marion Billaux