An antagonistic controller which is to be used in tendon-driven master/slave end-effectors is described. High stiffness was achieved by developing a control law which eliminates one local force loop in the flexion/extension control loops. One force loop maintains tendon cocontraction, while the other force loop is eliminated when the tendon tension is greater than cocontraction. In order to minimize the input impedance or slew drag caused by inertia and intrinsic damping of the system, force compensations were implemented according to the characteristics of the actuators. These techniques have been implemented experimentally on both electromechanical and electrohydraulic systems. A computer analysis of an electromechanical antagonistic control system is included. Servostiffness equations are derived from the transfer function of the electromechanical system.>

The effect of the variation of parameters of system elements on the overall performance of a generic model of a hydraulic actuation system for a robot is investigated. Specifically, an examination is made of the effects on actuator performance of two issues: intrinsic compliance i.e. the physical compliance within the actuator itself, and independent control of the actuator valve areas (e.g. supply and return areas for hydraulic fluid chamber) versus control of actuator valves with fixed area relationships. Increasing intrinsic compliance in the actuator degrades response to controller commands but improves the ability of the actuator to tolerate disturbances. Independent control of valve areas provides better response to commands and better rejection of disturbances than control with valves that have fixed area relationships. The performance information provided by the model permits behavior-based design of hydraulic actuation systems.>

Two antagonistic control algorithms are described. These algorithms are used to control manipulator links which are antagonistically driven by two actuators via tendons. They have been simulated and experimentally shown to produce better active and passive performance for an electric test system than control algorithms developed earlier. The algorithms use both positive and negative (push and pull) commands to be given to the actuators. Previous systems generated only pull commands, ensuring that tendons would not go slack and give rise to backlash and other problems. The proposed controllers allow push commands to the actuators but still do not allow tendons to go slack. Each actuator, in addition to being fed back its respective tendon force, is fed back both positive and negative manipulator joint torques. This feature allows both actuators to simultaneously respond to torque errors.>

The Center for Engineering Design at the University of Utah, and the Artificial Intelligence Laboratory at the Massachusetts Institute of Technology have developed a robotic end effector intended to function as a general purpose research tool for the study of machine dexterity. The high performance, multifingered hand will provide two important capabilities. First, it will permit the experimental investigation of basic concepts in manipulation theory, control system design and tactile sensing. Second, it will expand understanding required for the future design of physical machinery and will serve as a "test bed" for the development of tactile sensing systems. The paper includes: 1) a discussion of issues important to the development of manipulation machines; 2) general comments regarding design of the Utah/M.I.T. Dextrous Hand; and, 3) a detailed discussion of specific subsystems of the hand.