Neuromuscular Insights for the Design and Control of Hand-Assistive Robots

Abstract

Engineers often neglect the dynamics of neuromuscular control when developing hand-assistive robots. Models of the hand that decouple ‘peripheral’ kinematic and sensory dynamics from the ‘central’ dynamics of neural computation yield low-fidelity representations of the hand. This thesis presents approaches to model hand biomechanics and somatosensation and uncovers a candidate neuromotor solution manifold for neural control in coordinated hand function to help bridge central and peripheral dynamic models toward the development of robust hand-assistive robots. First, this thesis presents an integrated biomechanical-sensory testbed model of the hand featuring anthropomorphic passive joint impedances. Second, it presents a haptic design study that leverages this testbed model for artificial somatosensitivity and haptic evaluation. Finally, it identifies synergy-domain metrics for active joint impedance modulation to inform the development of neuromechanical models integrating peripheral and central dynamics. In the novel neuromuscular modeling approaches presented here, this thesis seeks to enable comprehensive design and control of hand-assistive robots.