Event Location: GHC 6501

Abstract: Many approaches to improving robotic manipulation have focused on reach-to-grasp tasks, where the arm motion and hand configuration are planned for grasping an object. In these solutions, the object placement is often considered fixed in the environment and is carefully grasped from its presented configuration. In contrast, humans often take advantage of an object’s movability to re-orient and re-grasp an object during the acquisition process.

This thesis investigates how such pre-grasp interaction can improve grasping through preparatory manipulation of the object’s configuration. Specifically we study the strategy of pre-grasp object rotation for grasp acquisition prior to a transport task. First, we examine the human performance of the pre-grasp rotation strategy for manual tasks. We find that the amount of pre-grasp object rotation was correlated with the change in the lifting capability of the grasping posture. In addition, the variability in the selected object orientation for the lift-off timepoint decreased for increased task difficulty based on increased object mass or increased task precision constraints. Second, we contribute a method for and evaluate the utility of planning pre-grasp rotation for a robot manipulator. Our results show that the pre-grasp rotation strategy can improve a robot’s manipulation capabilities by both extending the effective workspace for grasp acquisition and improving the quality of the transport action.

Poster

Committee:Nancy S. Pollard, Chair

Chris G. Atkeson

Jessica K. Hodgins

John M. Hollerbach, University of Utah