We present a new motion planning method for a biped robot. The method uses elevation map of the environment to plan the path of the feet and robot's body on rough terrain. We show how to incorporate modules which determine footholds, posture of the robot and check if the planned path is secure into a single general framework based on the Rapidly-exploring Random Trees. We describe the controller which stabilizes the robot during execution of the planned path. In the paper the implementation of the method for the Atlas robot in Gazebo environment is presented.
Mikolaj Wasielica, Dominik Belter, RRT-based motion planner and balance controller for a biped robot, Climbing and Walking Robots, CLAWAR 2016