Prospects for Combining Task and Motion Planning for Bi-Manual Solution of the Rubik’s Cube

We first describe a robot that solves the Rubik’s cube using separate task and motion planning with sensor-enabled closed-loop execution. The task solver uses iterative deepening Astar to plan the sequence of cube moves. Then a special-purpose grasp planner and controller places the manipulator hands on the cube to execute the task plan. The task requires two hands and two-handed re-grasping maneuvers. We then discuss the potential benefits of combining the task and motion planning, and discuss potential approaches to making this combination. Finally, we consider the prospects for combining task planning, motion planning, and sensor-based closed loop control.