A novel trajectory planning approach for obstacle-crossing based on contour-following nurbs for legged robots

Abstract

Legged robots possess superior terrain adaptability, and the planning of foot trajectories is directly related to their mobility. Confronting with complex uneven terrains and the limited single-leg workspace, existing foot-end trajectory planning methods struggle to ensure terrain adaptability while balancing motion smoothness and real-time performance. In this article, a novel trajectory planning approach based on contour-following NURBS (Non-Uniform Rational B-Splines) and convex envelope is proposed to achieve real-time obstacle-crossing walking for legged robots. By performing a convex envelope on the collected obstacle point cloud, the obstacle polygon to be simulated is obtained. The non-uniformity of NURBS is utilized in a rapid trajectory adjustment algorithm to generate a highly adaptive obstacle-crossing trajectory while ensuring smoothness. To verify the feasibility of the proposed method, both simulation and prototype experiments of the legged robot traversing non-uniform terrain are conducted. It is proved that the proposed method is a fast, effective and universal foot trajectory planning method suitable for uneven terrains, which can significantly enhance the obstacle-crossing capability of legged robots.