Towards a Parameterizable Exoskeleton for Training of Hand Function After Stroke

This paper describes the mechanical design, actuation and sensing of an exoskeleton for hand function training after stroke. The frame is 3D-printed in one piece including the joints. Apart from saving assembly time, this enables parametrization of the link sizes in order to adapt it to the patient's hand and reduce joint misalignment. The joint angles are determined using Hall effect sensors. They measure the change of the magnetic field of in the joints integrated magnets achieving an average accuracy of 1.25 degrees. Tendons attached to the finger tips transmit forces from motors. The armature current, which is proportional to the force transmitting tendons is measured using a shunt and controlled by a custom-made current-limiter circuit. Preliminary experiments with a force/torque-sensor showed high linearity and accuracy with a root mean square error of 0.5937 N in comparison to the corresponding forces derived from the motor torque constant.