Objective To realize the automatic walk and uniform application of mobile robot in the greenhouse, a kind of spraying robot capable of navigating autonomously between crop lines was designed.
Method Aiming at the problem that the recognition of navigation path was greatly affected by light changes, HIS space was selected from the color images acquired by Kinect camera. The clustering center and number were optimized for K-means algorithm. Using the improved K-means algorithm, the components of H and S were segmented and the complete road information was obtained. The method of Candy operator was used to detect the edge, and the method of improved Hough change was used to fit the navigation path. The method of fuzzy control was used to correct robot walking offset by adjusting the rotating angle and turn in real time. A self-tuning fuzzy PID control algorithm was selected for this spraying system to meet application requirements of different crops.
Result The system could effectively adapt to different light conditions. On average, it took 12.36 ms to extract the center line of crop. The navigation deviation does not exceed 5 cm. The coverage rate of plant leaves on upper, middle and lower layers was 63.26%, 50.89% and 75.82% respectively, and the droplet number per square centimeter was 55, 42 and 78 respectively.
Conclusion This system can meet the need of pesticide application of mobile robot to prevent pests and diseases in greenhouse.
DownLoad: