We are building an anatomically-correct testbed (ACT) hand to serve in three capacities: a) As a telemanipulator that mimics both the active and passive dynamics of a human hand for precision teleoperation and prosthetics, b) As an experimental testbed to investigate the complex neural control of human hand movements, and c) As a working physical model of the human hand for neuro- and plastic-surgeons to test new surgical reconstruction techniques for impaired hands. Here is a video demonstrating ACT Hand feature: ACTHand_Video

This hand, unlike many anthropomorphic robotic hands, incorporates biomechanical aspects of the anatomy that are functionally crucial in order to use control signals that resemble the neural commands. For example, the tendon insertion points and the routing of the tendons including the extensor mechanism (an intricate web-like tendon structure that sheathes the finger) are preserved. We identified the importance of the extensor mechanisms and its ability to provide independent control of the metacarpophalangeal (MCP) joint and acts not only as an extensor but also as a flexor, abductor, adductor, or rotator depending on the finger's posture.

The bones are machined from human bone data, and are accurate in surface shape, mass, and center-of-gravity, while joints have been designed to match both degrees-of-freedom and passive stiffness. Our evaluation of the assembled index finger confirms the anatomic properties, and reveals the function of some of the peculiar shapes of the finger bones, necessity of matching the joint passive stiffness properties, and connections of the extensor mechanism.

The actuators for the hand mimic both the active and passive behaviors of the human muscle. We use custom designed brushless DC motors that simulate active contraction. The motors connect to the tendons and apply torques on the finger joints to achieve desired position and end point force. A custom-made spring composite was used to simulate the human's nonlinear passive muscle stiffness closely. In order to study the neural control of hand movements, we are developing algorithms based on the muscle models to imitate muscle control using the DC motor. The goal is to be able to specify the muscle activation level, and the appropriate tendon tension could be provided given the current muscle length.