Online, Interactive User Guidance for High-dimensional, Constrained Motion Planning

Abstract

We consider the problem of planning a collision-free path for a high-dimensional robot. Specifically, we suggest a planning framework where a motion-planning algorithm can obtain guidance from a user. In contrast to existing approaches, we suggest to seek user guidance only when the planner identifies that it ceases to make significant progress towards the goal. User guidance is given in the form of an intermediate configuration $\hat{q}$ which, in turn, is used to bias the planner to go through $\hat{q}$. We demonstrate our approach for the case where the planning algorithm is Multi-Heuristic A* (MHA*) and the robot is a 34-DOF humanoid. We show that using this general approach allows to compute highly-constrained paths such as climbing stairs with little domain knowledge. Without our approach, solving such problems require carefully-crafted domain-dependent heuristics.