Footstep planning for humanoid robots among obstacles of various types

Abstract

The unique ability of humanoid robots to step over or upon obstacles is left unexploited if ordinary mobile robot navigation strategies are used for humanoids as well. Recently presented path planning strategies that make use of this capability, however, are very time consuming due to high computational complexity. We have presented a novel approach to humanoid robot footstep planning in obstacle cluttered environments that employs a human-like approach to terrain traversal. This paper mainly centers on providing a proof of concept for the algorithm while also bringing about some improvements. Unlike previous statically stable simulation work, here we present footstep plans including dynamic walk together with practical results of HRP-2 humanoid robot navigating in various scenarios with different obstacle characteristics. These also include catering for body sway dynamics as well as planning and executing stepping over obstacles encountered at angles not perpendicular to the robot's line of motion.