Robot navigation in vineyards based on the visual vanish point concept

Abstract

One of the biggest challenges of autonomous navigation in robots for agriculture is the path following in a large dimension map and various terrains. An important ability is to follow corridors and or vine rows which are frequent situation and with some complexity given the outline of real vegetation. One method to locate and guide the robot in between vineyards is making use of vanishing point detection on vine rows in order to obtain a reference point and send the adequate velocity commands to the motors. This detection will be conceived utilizing convectional image processing algorithms and Deep Learning techniques. It will be necessary to adapt the image processing algorithms or Deep Learning for use in ROS 2 context.