Research On The Key Technology Of Autonomous Navigation For Orchard Mobile Robot

Currently,fruit-growing industry is relying heavily on human labor,limiting the development of the industry, and industry upgrading is urgently needed.Automatic navigation technology for orchard mobile robot is a forward-looking technology in the automated and intelligent production of the orchard, which is of great significance to the scale and standardized operation of the orchard in the future.According to the characteristics of the environment of orchard, this paper studies the key technology of autonomous navigation for orchard mobile robot,with a comprehensive application for technologies such as image processing, intelligent control and sensor data processing.The key to realize orchard mobile robot's autonomous navigation is how to build the environment map of the orchard and to locate the robot with the map and all kinds of sensor data.This paper take the self-designed orchard mobile robot platform as a carrier to research, and derives the kinematics equations of the mobile robot platform's dead reckoning and motion control by analyzing the kinematics model. In view of orchard's planting by line according to standard specifications, this paper identify the tree, extract the coordinate information of the tree from the aerial view of the orchard using the methods of image processing such as threshold segmentation, edge detection and centroid extraction, and then create the characteristic map of the orchard. Using the lidar sensor to obtain the point cloud data of trees, it identifies the line characteristics of trees using the two line extraction algorithms Hough transform and random sample consensus( RANSAC),then compares the two algorithms with the use of simulation.By constructing the prediction models based on dead reckoning and the measurement model baesd on testing orchard's point features and line features and matching the map, the extended kalman filter algorithm is transplanted to the self-localization system of the orchard mobile robot. Based on localization results, the state flow of robot's autonomous navigation in the orchard is designed. Finally, this paper verifies the feasibility of the positioning algorithm and the autonomous navigation strategy through simulation and the field experiment.The simulation and experimental results show that, the map construction method based on image processing in this paper can be successfully used to generate the feature map of the orchard; The two line extraction algorithm can both accurately extract the tree line,but random sample consensus(RANSAC) has better real-time performance; Location algorithm with extended Kalman filter can help the robot locate itself,and autonomous navigation in the orchard is realized.