Terrain-Aware Trajectory Optimization for Autonomous Vehicle in Off-Road Terrain

Engineering notes

Simulation studies demonstrate that the proposed framework outperforms baseline planners in terms of safety and efficiency.

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

Autonomous navigation in off-road environments is critical for applications such as exploration, search and rescue, etc., where structured roads are unavailable. These scenarios demand effective trajectory optimization algorithms that account for sensor uncertainty, terrain geometry, and vehi-cle dynamics. However, most existing methods either overlook the uncertainty or simplify the interactive relationship between the vehicle and the terrain. As a result, the generated motion plan tends to be overly conservative and struggles to balance the tradeoff between robustness and computational efficiency. This paper presents a terrain-aware motion planning method that incorporates chance-constrained trajectory optimization to ensure a maximum probability of safe traversal. Unlike prior works that treat terrain geometry and dynamics con-straints separately, our probabilistic traversability index ex-plicitly couples vehicle state uncertainty which enables effective identification of traversable regions. To solve the optimization problem efficiently, a scenario-based method is adopted. Relying on the representative traversability samples along the nominal trajectory, this approach avoids excessive sampling, thereby achieving fast convergence while maintaining robustness to sensor uncertainty. Simulation studies demonstrate that the proposed framework outperforms baseline planners in terms of safety and efficiency.

Links and sources

• Official / arXiv page

Comments