Heuristic Search for Task and Motion Planning

Manipulation problems involving many objects present substantial challenges for motion planning algorithms due to the high-dimensionality and multi-modality of the search space. Symbolic task planners, on the other hand, can efficiently construct plans involving many entities but cannot incorporate the constraints from geometry and kinematics. In recent years, there have been a number of approaches proposed that integrate symbolic task planning with motion planning to attempt to get the best of both worlds. Many of these proposals have attempted to use planners, whether task or motion, as black-boxes, resulting in systems with weak heuristic guidance. In this paper, we show how to generalize the ideas behind one of the most successful symbolic planners in recent years, the FastForward (FF) planner, to the motion planning context. In particular, we show how to incorporate reachability in a configuration-space roadmap into the FF heuristic. The resulting tightlyintegrated planner is simple and efficient in a set of rep-