Receding-Horizon Path Planning for Risk-Free Mapless Navigation in Uneven Terrains
Yinchuan Wang, Xiang Zhang, Yuhan Wang, Chaoqun Wang, Rui Song, Yibin Li · IEEE/ASME Transactions on Mechatronics · 2025
It is an intractable question for the autonomous robot to navigate in uneven terrain, especially without a prior map of the environment. In this situation, the robot must judge where to go and maintain self-safety together, which is challenging for path planning and decision-making module of the robot. To address this problem, this study presents an integrated navigation framework for leading the robot to traverse unknown and uneven terrain and reach a designated goal. We incrementally construct a risk-aware tree during the mapless navigation. The tree nodes are growing on the ground surface while the edges reflect the risk and safety of traversing this edge. The root of the tree moves following robot position for rapid path finding. In addition, we design a receding-horizon strategy for determining where to go in the unknown environment. The valuable tree nodes located at the boundary of the local grid map are filtered to select a series of subgoals. These subgoals guide the robot to reach the final target step by step, and the paths to the subgoals are always optimized by rewiring the risk-aware tree for efficiency and safety. We evaluate our framework in both simulation and real-world experiments. The experimental results show that our approach has higher efficiency and efficacy compared with state-of-the-art methods.