Real-time path planning for humanoid robot navigation

Abstract

We present a data structure and an algorithm for real-time path planning of a humanoid robot. Due to the many degrees of freedom, the robots shape and available actions are approximated for finding solutions efficiently. The resulting 3 dimensional configuration space is searched by the A * algorithm finding solutions in tenths of a second on lowperformance, embedded hardware. Experimental results demonstrate our solution for a robot in a world containing obstacles with different heights, stairs and a higher-level platform. 1