A combined visual navigation method for greenhouse spray robot

Abstract

In the greenhouse environment, in order to solve the problem that there be deviations in the straight path of the spraying robot, and the navigation information needs to be quickly obtained at the turn, a navigation algorithm combining the red navigation line and the QR code is designed. After processing the acquired navigation line image, the navigation correction processing is performed based on the obtained deviation and yaw angle. Thereby achieving precise navigation in straight lines.However, red navigation line has problems such as low turning efficiency and poor corrective effect of deviation during cornering. Therefore, a QR code navigation method was designed to accomplish the requirement of prompt and accurate navigation at a turning point by identifying the navigation commands stored in the QR code.A total of 120 navigation line images and 120 QR code images under various conditions were collected to verify the algorithm. The average correction rate of the navigation line is 93.33%, and the average recognition rate of the QR code is 92.50%. The result of the navigation experiment shows that the position deviation can be controlled at 2.2 cm under different navigation speed conditions. The average deviation angle of heading angle is kept below 2.1°, and the maximum deviation angle is 4.6°. Which meets the real-time and accuracy requirements of the navigation of spraying robots.