Derivation and verification of synergy coordinates for the DLR hand arm system

Dexterous robotic hands are still not used in production lines, mainly because of the complexity to operate such hands. Most often a robotic hand is used to grasp and hold the object of interest while the object motion is performed by the arm. From studies on human grasping it is known that most of the grasping task can be reduced to two independent degree of freedom which drive the joints of the hand. Based on this idea, we analyzed a robotic grasp database to find suitable robotic “synergy coordinates”. 77% of these grasps can be represented by two coordinates that were originally defined by the 19 joint variables of the DLR Hand Arm System. In this work we also discuss the importance of the grasp approach configurations, which have significant influence on the results. The synergy coordinates were used to extend the joint level impedance controller and to provide a complexity reduced interface to an user or a planning algorithm, which leads to simplified operation of the hand and to reduced teach-in-times. From a different perspective this concept allows to imitate the behavior of a synergistic, respectively underactuated, hand by means of programming. The controller was evaluated successfully on the DLR Hand Arm System by realizing several different grasps covering a simplified grasping taxonomy. The DLR Hand Arm System has additional degrees of freedom to adjust the mechanical joint stiffness, which was used to modify the grasp force at a given pinch grasp configuration.