Research On Path Tracking Control Method Of Unmanned Rice Transplanter

China is a large agricultural production country,with the steady development of urbanization and deepening of aging,young adults are less willing to engage in agricultural labor,and the labor force tends to secondary and tertiary industries,resulting in the lack of rural labor and high labor costs;in addition,the farming time is concentrated and the operation level of farm mechanics varies,as well as the scale of farm construction,the development of unmanned farm machinery is an inevitable trend.At present,the unmanned farm machinery operation covers all production aspects of tillage,seeding,management and harvesting,and the precise operation can reduce costs and increase efficiency,and the path tracking control algorithm directly affects the operation effect,so it is significant to carry out research on the path tracking control method.Among them,the pure pursuit method has few control parameters and high practicality,and the suitable foresight distance is the key and prerequisite for the tracking performance,but at present the foresight distance adaptive has not been researched to breakthrough.During the operation of agricultural machinery,due to the complex operating environment and spatial variability,the vehicle is prone to slip under unknown external interference,which affects the tracking effect.And a stable and reliable control system is the prerequisite to realize the verification control method and the guarantee of accurate and efficient operation.Therefore,in view of the above problem analysis,the original unmanned rice transplanter control system is improved and designed,and the research is carried out around solving the core problems such as the pure pursuit algorithm forward-looking distance cannot be adaptive and the existence of external disturbance in the tracking control.The main research contents are as follows.(1)Improved design of software and hardware platform of agricultural machine control system.The hardware platform of the original unmanned rice transplanter control system was improved,including the front wheel turning angle measurement system,the steering control system,and the underlying control hardware.In order to meet the high-precision measurement of steering angle and realize the real-time position and status judgment of steering motor,a wheel vehicle active wheel turning angle measuring device and a gear box with built-in multi-turn absolute encoder were designed respectively;in order to improve the integration and reliability,the bottom control hardware improvement was carried out,including remote control system,bottom controller,adding serial screen,etc.Software development was also carried out,including source program transplantation and new functional subroutines.(2)Research on pure pursuit path tracking control method based on PSO-RBFNN.A pure pursuit control method based on PSO-RBFNN is proposed for the problem that the forward-looking distance is not adaptive.Through 276 sets of concrete pavement tests,1104 sets of high-quality samples are obtained by Latin hypercube sampling;with absolute value of deviation degree and speed as input and forward-looking distance as output,PSO-optimized RBFNN key parameters are trained on 904 training sets and 200 testing sets are predicted based on the training model,and the MSE of PSO-RBFNN model is 0.0033 m,which is better than 0.0074 m and 0.0060 m for the GD-RBFNN and BPNN models;the optimal parameters of the derived RBFNN are solidified in the underlying controller.In order to verify the effectiveness of the adaptive forwardlooking distance,three sets of concrete pavement surface comparison tests are established;and the results showed that the RMSE of the overall lateral deviation was3.17 cm and the MAE was 2.67 cm by water field operation tests.(3)Research on improved terminal slip mode control method based on interference observer.For the unknown disturbances in the actual operating environment,an extended kinematic model containing slip is established,based on which the composite disturbances in lateral and heading directions are introduced,a nonlinear disturbance observer is designed to avoid jitter,and a controller designed with a fast non-singular terminal slip mode surface and a variable exponential power convergence law is proposed to weaken the disturbances.Comparative simulations of the model without and with disturbances are performed in MATLAB simulation platform to demonstrate the feasibility of the method.To verify the robustness and reliability of the method,comparative path tracking tests were conducted for wet and slippery concrete surfaces,loose land and paddy field environments;the results showed that the RMSE of overall lateral deviation was 2.91 cm and the MAE was 2.38 cm through a 400 m linear path paddy field operation test.