Home The Problem Our Approach Videos Downloads Contact

Using LiDAR data for the autonomous movement

of an agricultural robot throughout a field.

Stage de recherche - Master 2 SDS ISTIA 2016/2017

Flávio Barros Pimentel Malavazi

Dr. Rémy Guyonneau

Dr. Sébastien Lagrange

Franck Mercier


Abstract

This work represents a step towards reaching a fully automated agricultural robot, replacing the current existing methods of path finding for an algorithm centered at a variant of the PEARL method for energy based multi-model fitting. The method has also been allied with a filter for it's results and a data-filter that eliminates points that are too far to be considered part of the model. Other sensors were combined with the LiDAR readings in order to increase the program's reliability and allow control in scenarios such as the row-change, where there are no readings provided by the optical sensor. The results are encouraging and were compared against other methods, having better performance than those.

Key Words : Agricultural robot, autonomous navigation, RANSAC, PEARL, IMU, Crop Navigation, LiDAR Measurements, Line extraction

Path followed by the robot OZ Weeding Robot Manouvering sequence