Task-driven Support Polygon Reshaping for Humanoids

In this paper we address a task-driven motion generation method that allows a humanoid robot to make whole-body motions including support polygon reshaping to achieve the given tasks. In the proposed method, generalized inverse kinematics (IK) is employed with floating-base to generate humanoid whole-body motions that enable the robot to accomplish the tasks according to given priorities. During the motion, several criteria such as manipulability or end-effector position error are tracked. If the desired task cannot be done because of those criteria, a geometric planner for support polygon is activated. Then the whole-body motion is computed again always using the generalized IK solver including stepping motion that realizes the deformed support polygon by using dynamic walking pattern generator. This method provides a way to perform tasks that could not be done without changing the humanoid's support state. We have verified the proposed framework through simulations and experiments using humanoid robot HRP-2