Search

In this paper we present our low-level software framework that allows human neuromuscular properties and behaviors to be programmed modularly onto biologically inspired prosthetic devices. Specifically this software framework is used at the lowest level of the Anatomically Correct Testbed (ACT) Hand. It abstracts away from the robotic details and allows users including biologists to program individual muscle physiology, multiple muscle synergistic activities, neural commands, etc. independently. Our system has been designed to be fault-tolerant with the idea of safe and continued operation of prostheses if a new prototype behavior were to perform incorrectly or outright fail. Example coding of Hills model is shown.

Three human subjects participated in a closed-loop brain computer interface cursor control experiment mediated by implanted subdural electrocorticographic arrays. The paradigm consisted of several stages: baseline recording, hand and tongue motor tasks as the basis for feature selection, two closed-loop one-dimensional feedback experiments with each of these features, and a two-dimensional feedback experiment using both of the features simultaneously. The two selected features were simple channel and frequency band combinations associated with change during hand and tongue movement. Inter-feature correlation and cross-correlation between features during different epochs of each task were quantified for each stage of the experiment. Our anecdotal, three subject, result suggests that while high correlation between horizontal and vertical control signal can initially preclude successful two-dimensional cursor control, a feedback-based learning strategy can be successfully employed by the subject to overcome this limitation and progressively decorrelate these control signals.