Goal-contact relaxation graphs for contact-based fine motion planning

Abstract

In many robotic assembly tasks, fine motions are contact motions among parts which are inevitable because of uncertainties and are even desired to reduce uncertainties. Knowledge of the related contact states is extremely important in planning such motions. Representing contact states as contact formations based on the notion of principal contacts, this paper introduces goal-contact relaxation graphs (GCR graphs) to represent and organize neighboring contact formations of typical "goal" contact formations. The GCR graphs carry concisely the information essential to contact motion planning and can be generated automatically. A task-independent algorithm for generating major subgraphs of GCR graphs is given in this paper.