Simultaneous Planning of the Path and Supports of a Walking Robot

Abstract

In this paper we study the simultaneous planning of the path and leg supports of an eight-legged robot on uneven terrain. We use the A-star algorithm (A*), which searches for the shortest path between two points. First, the terrain is modelled with a triangular mesh and the triangles are subdivided to take the centroids of these triangles as the search space of the A*. Secondly, with respect to the original A*, the stability of the robot at each centroid is considered, so that the cost at a centroid is penalised if the robot is unstable (i.e., the robot slips and/or tips over), or the cost is zero if it is stable. The stability at each contact point is determined by calculating that the ground reaction at that point is contained in a linear approximation of the friction cone. Finally, the path, the contact points of each leg, as well as the robot's posture at each position are obtained