A Continuous Approach to Legged Locomotion Planning

Abstract

Legged locomotion planning can often be reduced to the search of finite sequences of contacts between the robot and the ground. We address this problem, using an original, continuous approach. We first demonstrate how this intrinsically discrete problem can be made continuous, and then present two applications: an experiment of vision-based reactive footstep planning with the DLR-Biped robot, and a simulation of non-gaited locomotion planning with a hexapod robot.