PDAC-Based 3-D Biped Walking Adapted to Rough Terrain Environment

This paper deals with the 3-D biped walking of a humanoid type robot over rough terrain. We previously proposed efficient 3-D biped walking control using Passive Dynamic Autonomous Control (PDAC) based on the assumption of point-contact and virtual holonomic constraint of robot joints. Walking adaptability has not, however, been analyzed. We thus analyze the environmental adaptability of PDAC-based walking method in this paper. The robot is modeled as a variable-length 3-D inverted pendulum whose dynamics is modeled as a 2-D autonomous system by applying PDAC. We analyze the stability of the 2-D autonomous system using a Poincaré map and derive the stable range of uneven height over rough terrain. We then experimentally validate 3-D biped walking on unknown rough terrain using our humanoid type robot, Gorilla Robot III.