Development Of Automatic Driving System For Rice Transplanter Based On GNSS And ROS

With the introduction of the concept of unmanned farm,agricultural machinery autopilot is one of the indispensable technologies in the future,and in-depth research on this technology is also being done at home and abroad,reflecting the significance and application value of its research.The use of agricultural machinery autopilot technology can improve the operation quality and efficiency on a large scale,and significantly reduce the production cost and labor intensity.In order to better realize the autopilot operation of agricultural machinery and meet the needs of the landing of unmanned farm projects in the future,the core technology that needs to be solved urgently is the research and development of autopilot system.The technology in this area is relatively mature abroad,but it started relatively late in China.At present,there are still some problems in the research field of self-driving technology in China: first,the domestic agricultural machinery navigation control system mainly uses middle and low-end microprocessors for bare metal control.Although it can achieve part of the functions of self-driving,in the long run,agricultural machinery autopilot systems need to integrate more sensors to achieve more intelligent operations.The computing power of embedded microprocessor alone is far from being able to meet the needs of autopilot.Second,at present,the front wheel steering angle measurement is still measured by the angle sensor,considering that the installation of the traditional connecting rod angle sensor is complex,which will have a great impact on the wheel angle measurement,thus affecting the stability of agricultural machinery autopilot.Third,taking into account the changes in the working speed of rice transplanter,paddy field side slip and other factors on the agricultural machinery path tracking control algorithm and so on.For this reason,this paper takes the rice transplanter as the research object,and the main research contents and achievements are as follows:(1)The hardware system of automatic driving operation of rice transplanter is built based on the architecture of upper and lower computer based on CAN bus(the upper computer is the high performance computing module and the lower computer is the controller).In terms of software,with the help of the ubuntu operating system of the highperformance computing module,the ROS robot operating system is built as a subsystem,and the autopilot function module node is designed according to the distributed architecture supported by ROS,and each node can run independently.Transplant some of the existing programs of the team to the ROS platform,design the upper and lower computer software architecture suitable for the system,communicate in the way of topic to establish the communication bridge between nodes,and greatly improve the research and development efficiency and computing ability of the system.(2)An information fusion estimation algorithm based on GNSS/MEMS gyroscope for measuring wheel angle is designed.The output measured by the body gyroscope and the wheel gyroscope is different,and the steering angle of the front wheel is calculated by integral.The two-wheel model of the rice transplanter is established,and the calculation formula of the front wheel angle is obtained according to the model.The double Kalman filter is used as the measurement equation of the filter to correct the zero deviation of the gyroscope when calculating the front wheel angle in real time.The experimental results show that when the rice transplanter runs in a straight line,the filtering effect is good,and the average error of the front wheel angle after filtering is 0.18 °,which is 1.08 °lower than that before filtering.(3)The method based on LQR optimal controller is designed to control the path tracking of the automatic driving operation of rice transplanter.The kinematics model of the twowheeled vehicle of the rice transplanter is established,and the formulas of lateral deviation and heading deviation are calculated from the model and used as the input of the LQR optimal controller.In order to make the objective function J reach the minimum value,it is necessary to select the most suitable weighting matrix Q and R to get the most suitable state feedback control parameter K.according to the model and the optimal control law,a simulation system is designed.The most suitable Q and R values are determined by simulating and observing the rising time,convergence time and convergence degree of the curve.After designing the LQR optimal controller,the linear path tracking test is carried out to analyze the path tracking performance of the rice transplanter at different speeds and different road conditions.Based on the above research results,the functional test of the rice transplanter autopilot system was carried out in the Ningxi Scientific Research and Experimental Base of South China Agricultural University,and the test results further verified the feasibility and reliability of the research results.