Path tracking method of soil sampling platform based on MPC-Stanley

To address the limitations of poor path tracking performance and low tracking accuracy during the autonomous navigation of soil sampling platforms, this study proposes a path tracking method for soil sampling platforms based on MPC-Stanley. Firstly, a navigation system hardware architecture was designed for the soil sampling platform, and a control scheme for the navigation system was formulated. Secondly, a kinematic model was constructed using a bicycle model for kinematic modeling. Subsequently, model predictive control (MPC) was selected as the path tracking controller, deriving the control processes for both the MPC and Stanley controllers. To address the MPC controller’s high computational complexity and difficulty in solving within specified time constraints, the Stanley controller was employed to optimize the MPC controller’s steering angle. The steering angle calculated by the Stanley controller serves as input to the MPC controller. To meet the real-time control requirements of the navigation system, the control quantities were modified using an exponential form, achieving model simplification. Finally, using a soil sampling platform as the control object, an inertial measurement unit (IMU) and satellite positioning module were employed to obtain real-time pose information of the platform, conducting field path tracking experiments. Using straight path T_r1 and curved path T_r2 as reference paths, tracking performance tests were conducted at a platform speed of 0.8 m/s for the MPC-Stanley controller, pure pursuit (PP) controller, and proportional integral derivative (PID) controller on both straight and curved trajectories. The tests demonstrated that the MPC-Stanley controller achieved the best path tracking performance. By predicting the system’s state variables based on the motion model, the MPC-Stanley controller effectively resolved the overshoot and oscillation issues observed in the PID controller, exhibiting superior robustness compared to the PP controller. Subsequently, tracking error tests were conducted for the MPC-Stanley controller, PP controller, and PID controller on the straight trajectory T_r1 and curved trajectory T_r2 at four speeds: 0.8, 1.6, 2.4, and 3.2 m/s. Test results indicate that during T_r1 straight-line path tracking, the MPC-Stanley controller achieved average absolute deviation, maximum absolute deviation, and standard deviation of 3.1 cm, 4.2 cm, and 1.2 cm respectively across all four speeds. Compared to the PP controller, these values improved by 43.6%, 43.4%, and 14.3% respectively. Compared to the PID controller, improvements were 20.5%, 23.0%, and 7.7% respectively. The MPC-Stanley controller fully leveraged the advantages of the dynamic model, avoiding the large system errors caused by the PP controller’s lack of angular control. Compared to the PID controller, the MPC-Stanley controller demonstrated more stable performance during path T_r1 tracking. During curved path T_r2 tracking, the MPC-Stanley controller achieved average absolute deviation, maximum absolute deviation, and standard deviation of 3.9 cm, 6.6 cm, and 1.5 cm respectively across four velocity conditions. This represents improvements of 80.2%, 79.8%, and 85.7% over the PP controller, and 93.0%, 89.8%, and 90.5% over the PID controller. The MPC-Stanley controller enables adaptive parameter tuning based on system inputs and outputs, delivering superior control performance. The path-tracking method designed in this paper effectively enhances the tracking accuracy of the soil sampling platform. Furthermore, this method is universally applicable to various agricultural operation platforms with structures similar to the soil sampling platform described herein, providing a technical reference for high-precision navigation operations in the field.