Autonomous Navigation with Human Observation for a Biped Robot

Abstract

To make a robot navigation according to the observation of human beings and fully realize its motion capability, a biped robot must be able to move with an efficient path and collision-free footstep in various three-dimensional (3D) environments. Thus, in this paper, a search strategy with distributed path planning and footstep planning is adopted to obtain a 3D path and a human-like set of footsteps in a 3D environment. For path planning, a Three-Dimnesional for Biped Robot A* (TDB-A*) algorithm based on the A* is proposed to plan an optimal and safe path considering human tendency factors, robot structure limit, walking ability and environmental constraints. The human tendency factors are quantified by human observations. For footstep planning, the towing points of the left and right feet are captured from the path on the basis of human behavioral habits and stability constraints, and then calculate collision-free and efficient footsetps to guide the biped robot. The validity of the proposed method for navigation over 3D environments was investigated by simulation.