Manipulation planning under changing external forces

This paper presents a planner that enables robots to manipulate objects under changing external forces. Particularly, we focus on the scenario where a human applies a sequence of forceful operations, e.g. cutting and drilling, on an object that is held by a robot. The planner produces an efficient manipulation plan by choosing stable grasps on the object, by intelligently deciding when the robot should change its grasp on the object as the external forces change, and by choosing subsequent grasps such that they minimize the number of regrasps required in the long-term. Furthermore, as it switches from one grasp to the other, the planner solves the bimanual regrasping in the air by using an alternating sequence of bimanual and unimanual grasps. We also present a conic formulation to address force uncertainties inherent in human-applied external forces, using which the planner can robustly assess the stability of a grasp configuration without sacrificing planning efficiency. We provide a planner implementation on a dual-arm robot and present a variety of simulated and real human-robot experiments to show the performance of our planner.