A novel vision based row guidance approach for navigation of agricultural mobile robots in orchards

This paper presents a novel vision based technique for navigation of agricultural mobile robots in orchards. In this technique, the captured color image is clustered by mean-shift algorithm, then a novel classification technique based on graph partitioning theory classifies clustered image into defined classes including terrain, trees and sky. Then, Hough transform is applied to extract the features required to define desired central path for robot navigation in orchard rows. Finally using this technique, mobile robot can change and improve its direction with respect to desired path. The results show this technique classifies an orchard image properly into defined elements and produces optimal path for mobile robot.