Planning of object motion by graspless manipulation using contact state transition graph

Abstract

We propose "contact-state transition graph" for planning of graspless manipulation. Graspless manipulation is human-like dexterous manipulation method in which the robot fingers do not grasp an object but manipulate it with a support by environment. In graspless manipulation, the object contacts with the environment in several ways. Several patterns of the contact state between an object and environment are described as nodes of graph and are connected by arcs which describe graspless manipulation methods. From this graph, we can find the minimum cost path from a start state to goal state. For a correct cost and state transition, we introduce a parameterized node in the graph.