Foothold-Based Planning for Legged Robot Autonomous Navigation Over Uneven Terrain
Jiyu Yu, Dongqi Wang, Zhenghan Chen, Ci Chen, Rong Xiong, Yue Wang · IEEE/ASME Transactions on Mechatronics · 2025
Legged robots hold promise for traversing challenging terrains, which is crucial for tasks, such as emergency response and inspections. Although the locomotion capabilities of legged robots have greatly improved, navigating unknown and uneven environments in real-time applications remains a key challenge. This is because existing controllers typically have a short prediction horizon and often require manual remote control. In this article, we propose a fast, foothold-aware path planner that enables rapid planning while simultaneously taking into account the feasibility of footholds along the path, and an online foothold and trajectory planner that plans appropriate footholds and collision-free whole-body trajectories in real-time. This layered approach does not require the long offline computation time, allowing for both real-time and long-horizon planning. We then integrate these planners with existing controllers and the perception module to achieve fully autonomous navigation of legged robots over uneven and unknown terrains. Extensive simulations and real-world experiments validate the effectiveness of our approach, demonstrating enhanced performance in navigating highly challenging terrain scenarios.