Slip-Aware Model Predictive Path Tracking Control for Tracked Agricultural Vehicles in Hilly Wheat Fields

Tracked agricultural vehicles operating autonomously in hilly wheat fields with small and fragmented plots are highly susceptible to track slip during rotary tillage operations, due to uneven terrain, frequent headland turning, and high traction loads. Such slip can significantly degrade path-tracking accuracy, leading to missed tillage, excessive overlap, and degraded tillage quality. To address these challenges, this study proposes a slip-aware path tracking control method based on model predictive control (MPC), integrating a sliding mode observer (SMO), adaptive modeling, and dynamic prediction horizon adjustment. Specifically, compound constraints and a velocity variation penalty term are incorporated into the MPC framework to enhance control smoothness and stability under high-slip conditions. A sliding mode observer is designed to estimate the left and right track slip ratios in real time, which are embedded into a nonlinear kinematic model to construct a slip-aware control system. Simulation results demonstrate that the proposed slip estimation method reduces the average lateral tracking error by 56.94% compared to fixed slip-compensation models, thereby significantly improving tillage trajectory consistency. Furthermore, a fuzzy logic controller is developed to dynamically adjust the MPC prediction horizon according to path curvature and vehicle speed, achieving a balance between tracking accuracy and computational efficiency. Field experiments confirm that, compared with a fixed prediction horizon, the proposed method reduces path tracking error and computational load by 15.22% and 30.29%, respectively, while increasing the effective working coverage rate to 96.8%.respectively, demonstrating its practicality for autonomous rotary tillage operations in hilly wheat field environments.