This research addresses the design and control of machines that will have stable dynamic interaction with humans via simultaneous exchange of both power and information signals. Typically, human interaction with machines involves an exchange of information signals only, i.e., pushing the start button on and activating an electric mixer, but not transferring power directly to the mixer, or feeling the load on the mixer blades. Orthoses, devices attached directly to a human for the purpose of performing a task or aiding a handicapped individual, are examples of machines which do involve the exchange of power and information between human and machine. This research is a five phase activity which focuses on (1) development of a learning controller for on line computer modeling of human hand dynamic behavior, (2) development of the rules for the design of a supervisor controller to create a particular behavior that humans cannot perform unaided, (3) study of the design of a multi degree of freedom orthosis that can exchange power and information signals with humans, (4) design and fabrication of a prototype orthosis and computing environment, and (5) experimental evaluation of the performance of this system. The ultimate objective is a prototype orthosis which transfers to the human's hand, in feedback fashion, a scaled down version of the actual external load which the orthosis is manipulating. This natural feedback force on the human's hand lets her feel a modified version of the external forces on the orthosis.
