Technology For Dexterous Remote Manipulation

Abstract

Remote manipulation of objects can provide for safer construction and operations in extreme environments such as space, underwater, and disaster areas. For manipulation of objects designed for human use such as levers, handles, knobs, and so forth, a dexterous hand is desirable. Robots can be operated in remote locations as telerobots with supervision by the human operator but with a certain degree of local autonomy. This requires some intelligence at the remote site. We describe recent work related to intelligent grasping by a dexterous mechanical hand. In order to provide robot hands with intelligence required for effective multifingered grasping, a strategy of grasp reasoning is being developed to plan the grasp so that the fingers will be guaranteed to firmly grasp the object during handling and generate fine motion to perform tasks using the same planned grasp points. Grasp planning strategies based upon heuristic experiences from the motions of human hands and fingers may experience difficulties when applied to the grasp of mechanical hands and fingers due to lack of precise sensing devices. This paper proposes a planned strategy for grasping based on mechanics principles and codes the grasp mechanics into an expert system which has been applied in graphic simulation. This system can be used to reason and make decisions of grasp points on any polygonal object. This strategy has the advantages of being intelligence based and computationally efficient without sacrificing the physics of the grasp. The method will be useful in real-time applications of sophisticated robot hands by reducing the complexity both in planning and control. This system cm have applications to the grasping of objects for construction and operation by mechanical hands with semi-autonomous operations in extreme environments.