Research Of Autonomous Driving System Of Tractor Based On Multi-source Information Fusion In Orchard

Achieving the autonomous driving of the tractor in orchard is an important way to improve the productivity of China's fruit industry,and to develop modern agriculture.The autonomous driving system contains a variety of elements,the most important of which include perception-location,decision-making control and execution.The tractor's autonomous driving system in orchard is designed by focusing the three key parts.(1)Design and construct the software and hardware platforms for the tractor.The driving device attaches to the steering wheel of the tractor,and the steering is detected by directly measuring the front wheel angle.It guarantees the real-time and accuracy of the measurement.In order to make the operation of the system more reasonable and efficient,the software platform is divided into two parts: the upper part and the lower part,to fulfil the different requirements of the controller and the actuator.The upper system is based on ROS,and the lower system is based on the lightweight embedded system.(2)Research the perception-location method based on multi-source information fusion.In orchard,the fruit trees are selected as landmarks to construct the map of the orchard,and distinguished by the spatial characteristics of pointclouds in a new method.Research the application of EKF SLAM in orchard.Fuse the data of the lidar,GNSS and INS in the EKF SLAM system,to improve performance of the perception-location system in orchard.(3)Research the decision-making strategy for the orchard.According to the distribution of fruit trees in orchard,the path planning is divided into two parts including the global path planning and the inter-row path planning.The global path planning adopts the zigzag driving to reduce the accumulated error of the perception-location system.The inter-row path planning provides the desired path by using Hough transformation and the least-square method,to adapt the distribution of the fruit trees in the inter-row.The path tracking system is based on the pure pursuit model with lookahead distance adjusted by fuzzy control to solve out the desired front wheel angle,in order to optimize the path tracking of the tractor.In addition,feedforward control is added to the path tracking system to compensate the sideslip caused by the soft terrain and the vibration of tractor.(4)Test the autonomous driving system.Selecting the experimental scenario,the autonomous driving system is tested to analyze its performance and to verify the reliability and applicability of the system in orchard.