An efficient motion planning framework for four-wheel steering autonomous vehicles using Lazy Edge-Based A* and adaptive RK4-MPC

Engineering notes

The integration of adaptive RK4-MPC with WFDC achieves the lowest tracking error and heading error of 34.8 % and 37.5 % compared to OMNI, and 28.6 % compared to S-4WS. For computational efficiency, the proposed framework achieves a search time of 0.5234 s, 83.1 % faster than OMNI, and 37.1 % faster than S-4WS, and an optimization time of 1.4892 s, 30.3 % faster than S-4WS.

Chinese explanation / 中文解读

中文解读待补充：本站会优先为端到端自动驾驶、BEV感知、3D目标检测、轨迹预测、路径规划、LiDAR感知等高价值论文补充中文说明。

Original abstract

This work presents an efficient motion planning framework for four-wheel steering (4WS) autonomous vehicles operating in complex and unknown environments. To improve planning efficiency, the framework employs a lazy edge-based A* (LEA*) algorithm for global path planning, adaptive fourth-order Runge–Kutta model predictive control (RK4-MPC) for trajectory tracking and motion execution, and wheel force distribution control (WFDC) to ensure stable motion during steering maneuvers. Quantitative results show that the LEA* reduces planning time by 87.5 % edge evaluations by 96.1 % compared to conventional A*, while improving path smoothness by 51 %. The integration of adaptive RK4-MPC with WFDC achieves the lowest tracking error and heading error of 34.8 % and 37.5 % compared to OMNI, and 28.6 % compared to S-4WS. In addition, the proposed method reduces the wheel slip ratio 88.4 % better than OMNI and 46.7 % better than S-4WS, while also reducing yaw acceleration by 50 % compared to both baselines. For computational efficiency, the proposed framework achieves a search time of 0.5234 s, 83.1 % faster than OMNI, and 37.1 % faster than S-4WS, and an optimization time of 1.4892 s, 30.3 % faster than S-4WS. Overall, the proposed framework improves motion planning efficiency while maintaining smooth and stable motion in simulation.

Links and sources

Comments