Robust autonomous row-change maneuvers for agricultural robotic platform

In this work, we present a row-change maneuver algorithm based on lines detection from a LiDAR point-cloud. The control software has been tested on a tracked robotic platform while traversing different types of orchards, both in terms of tree shapes (planar, open vase, spindle) and rows distances (from 2m to 5m). The use of the LiDAR sensor is motivated by the need of both high precision in the maneuver (less than 15cm of tolerance while entering rows) and the capability of detecting an environment that is constantly changing, as it is the case when dealing with plants that follows a seasonal growth. Experimental tests are analyzed and results presented.