Visual Feedback Manipulation for Hand Rehabilitation in a Robotic Environment

The ultimate goal of this research is to manipulate visual feedback in order to make robotic therapy more effective than traditional human-assisted and robotic rehabilitation. Toward this end, this thesis examines the limits and effects of visual feedback distortion, a type of visual feedback manipulation, in a robotic environment. Visual feedback establishes a metric of performance for a given task, and visual feedback distortion is defined as a change in this metric such that a change in performance is required to cause the same visual response. In order to design a therapeutic program utilizing visual feedback distortion, it is crucial to identify the amount of distortion that is imperceptible. Thus, the first experiment measured the amount of distortion that is imperceptible (the Just Noticeable Difference, or JND) for age-matched, unimpaired subjects and motorimpaired subjects. The second experiment showed that vision dominates kinesthetic feedback in a robotic environment and that gradual visual distortion beyond 1 JND increases force production and movement distance within a single experimental session. Finally, the third experiment examined the effects of distortion of movement error during a two-finger coordination task. The results of this experiment showed that error distortion is effective in directing a subject’s attention to a specific finger but does not improve terminal performance. These foundational experimental results were used to design a game-like therapeutic paradigm incorporating visual feedback manipulation. In initial tests with three chronic stroke and traumatic brain injury patients, all patients followed the visual feedback manipulation to levels of performance above that predicted by their initial assessment at each session. Furthermore, all patients showed functional improvements after participation in the study. Visual feedback manipulation has shown promise for therapy in a robotic environment; more work is needed to further explore the ramifications of visual feedback manipulation for robotic rehabilitation and to spread this technique to clinical practice.