Autonomous Navigation of a Humanoid Robot Over Unknown Rough Terrain

This paper presents an integration of laser-based perception, footstep planning, and walking control of a humanoid robot for navigation on previously unknown rough terrain. Perception system that obtains the shape of surrounding environment in a few centimeters accuracy is realized by using scanning type laser range sensor as input. A footstep planner decides the sequence of stepping positions by using the obtained terrain shape. A walking controller that can manage a few centimeters error of the terrain shape measurement is achieved by a combination of 40 [ms] cycle online walking pattern generation and a sensor feedback ground reaction force controller. An operation interface that is developed for giving commands to the robot is also presented. Mixed-reality display is adopted for realizing intuitive interfaces. Navigation system is implemented on the full-size humanoid HRP-2. Several experimental results on unknown rough terrain shows the performance of the proposed system.