Foot-End Obstacle Avoidance Trajectory Planning for Quadruped Robots

摘要

Achieving dynamic locomotion of legged robots in rugged terrains requires precise control of foot placement and collision avoidance. Currently, some advanced motion controllers utilize external perception information for foot planning in such scenarios, with foot-end obstacle avoidance methods typically relying on constructing Signed Distance Function (SDF) maps. However, this approach has limitations in terms of real-time performance and computational overhead. To address this issue, this paper proposes a novel foot-end obstacle avoidance trajectory planning method based on elevation map information. This method eliminates the reliance on SDF maps and instead analyzes elevation map information to select optimal foothold points, compute collision gradients using local collision-free guidance paths, and generate smooth trajectories through a Cube B-spline curve optimizer. Compared to traditional methods that require the construction of SDF maps incurring computational costs, this approach significantly enhances computational efficiency.