Rice row tracking control of crawler tractor based on the satellite and visual integrated navigation

To achieve autonomous navigation of the crawler tractor in the paddy field, a set of control system was developed by integrating the satellite and visual navigation. The system is mainly composed of a satellite module, a visual module and a controller. The satellite module detects the deviation information between the tractor’s position and the pre-set line AB using GPS Real–time Kinematic (RTK) technology and the navigation is achieved according to the preview point tracking algorithm. The visual module obtains the rice crop images in real time by camera, and the binary image segmentation is completed by ExG(2G-R-B), Otsu and mask method. Then the region search algorithm is used to classify the region of each rice row, and the visual navigation path is fitted by the weighted least square method, and the weight value is based on the lookahead distance. Finally, the preview point tracking algorithm is used to realize the visual navigation. Furthermore, the control division line between the satellite navigation and visual navigation is established, the threshold value of which is determined based on the crop row deviation due to the satellite navigation when rice crops were planted. The crawler tractor navigation tests were conducted to evaluate the performance of the integrated navigation system. When the speed of the tractor was 0.4∼0.6 m/s, 0.7∼0.9 m/s and 1.0∼1.2 m/s, the maximum lateral deviation of the tracking by visual navigation was 0.18 m, 0.15 m and 0.14 m, respectively, and the RMSE was 55 mm, 64 mm and 60 mm, respectively. Under the same speed, when the threshold value was set to 0.15 m, the maximum lateral offset of the tracking by integrated navigation was 0.14 m, 0.13 m and 0.12 m, respectively, and the RMSE was 41 mm, 50 mm and 55 mm, respectively. The experimental results indicate that the satellite navigation module in the integrated navigation system functions well in rice row tracking when some seedlings are missing, and significantly reduces the navigation deviation and the degree of deviation fluctuation.