Autonomous Navigation Development of Solar Panel Cleaning Vehicle

This study presents an automated navigation system for solar panel cleaning vehicles. Since self-positioning and navigation cannot be achieved on solar panels through self-built mapping, we utilize the vanishing points formed by the perspective transformation of many coplanar parallel lines on the solar panel to accomplish navigation. This is complemented by fuzzy control to align the vehicle with the parallel lines and proceed along the straight lines of the solar panel. When the solar panel cleaning vehicle reaches the boundary, it will turn to the next row's direction to continue cleaning along the straight lines of the solar panel until the cleaning is complete or autonomous navigation is deactivated. Preliminary experimental results show that the average standard deviation of the x-coordinate during straight-line movement is 1.70 pixels, and the average error between the maximum and minimum x-coordinates is approximately 6.33 pixels. After adjusting the speeds of the two-wheel servo motors, calibrating the camera’s centre point, and repeating the experiments, the results demonstrate an average standard deviation of 0.43 pixels, and the maximum and minimum errors in the x-coordinate of straight lines have decreased to 2 pixels, achieving practical usability.