Studying the implementation of iterative impedance control for assistive hand rehabilitation using an exoskeleton

A positive training synergy can be obtained when two individuals attempt to learn the same motor task while mechanically coupled to one another. In this paper, we have studied how mimicking this interaction through impedance control can be exploited to improve assistance delivered by hand exoskeleton devices during rehabilitation. In this context, the machine and user take complementary roles akin to two coupled individuals. We present the derivation of a dynamic model of the human hand for the purpose of controller development for new hand exoskeleton platforms. Using this model, we have simulated the behavior of an iterative impedance controller programmed for rehabilitative training. The controller interacts with cylindrical objects to be grasped by means of an inverse kinematic mapping and tuning of mechanical impedance characteristic of the finger joints. Through fusion of concepts from motor control theory, muscle impedance and task oriented control, the controller is capable of iteratively learn to accomplish simple tasks involving grasping and lifting while cooperating with a user. The controller is also capable of adapting to more complex dynamics for more dexterous tasks, such as pulling on a hand-bar or loosening the cap of a jar. We believe the human-robot synergy established in this investigation has benefits to therapy. It can be combined with a broad range of training exercises and represents an incremental step towards mimicking natural human motor responses.