Machine Vision System for Autonomous Guidance of Raisin-Collection Rover

The development of autonomous vehicles based on machine vision system in various industries and agriculture is a challenging topic. In this study, the active-contour algorithm was utilized for path-tracking of an automatic raisin-collecting rover. For this purpose, the energy difference between the background and foreground levels is utilized to extract the boundary between raisins and the ground surface with good accuracy which is used for navigating the path. This was accomplished by developing a small unmanned rover equipped with four DC motors and a machine vision system. Field tests at the Sunny Raisin Court showed that the system was able to independently detect and navigate its path with an RMS error of 1.57 cm. The results of analysis of variance for accuracy of the path tracking showed that the effect of lighting conditions and speed on the accuracy was significant at the level of one percent.