Abstract:
The central theme in robotic manipulation is that of the robot interacting with the world through physical contact. We tend to describe that physical contact using specific words that capture the nature of the contact and the action, such as grasp, roll, pivot, push, pull, tilt, close, open etc. We refer to these situation-specific actions as manipulation primitives. Due to the nonlinear and nonsmooth nature of physical interaction, roboticists have devoted significant efforts towards studying individual manipulation primitives.

This thesis begins with the study of pulling and pushing primitives. First, we derive exact bounds on the motion of a pulling object and use these bounds to plan pulling trajectories which guarantee convergence. Second, we show that large-scale multi-object rearrangement is achievable using only simple pushing primitives. However, in spite of the close relation between pushing and pulling, the required engineering effort within each project limited our ability to generalize across projects or to new scenarios.

This limitation motivates the main contribution of this thesis. We contribute a complete and general framework to autogenerate manipulation primitives. To do so, we develop the theory and computation of contact modes as a means to classify and enumerate manipulation primitives. The contact modes form a graph, specifically a lattice. Our algorithm to autogenerate manipulation primitives performs graph-search on the contact mode lattice and solves a quadratic program to generate each primitive. We hope that our contributions will lead to more general approaches for robotic manipulation.

More Information

Thesis Committee Members:
Matthew T. Mason, Chair
Chris Atkeson
Aaron Johnson
Kevin Lynch, Northwestern University