Hierarchical planning for multi-contact non-prehensile manipulation

Abstract

Manipulation planning involves planning the combined motion of objects in the environment as well as the robot motions to achieve them. Ideally, we would like to be able to plan prehensile and non-prehensile manipulation in an integrated fashion, but despite much progress in planning motions in the robot's configuration space in the presence of obstacles, progress in general manipulation planning has been more limited. The manipulation planning problem has a number of complicating factors, including the large dimensionality of the combined space. In this thesis, we explore a hierarchical approach to planning sequences of non-prehensile and prehensile actions. We subdivide the planning problem in three stages (object contacts, object poses and robot contacts) and thereby reduce the size of search space that is explored. We show that this approach is more efficient than earlier strategies that search in the combined robot-object configuration space directly.