Impedance and force-field control of the index finger module of a hand exoskeleton for rehabilitation

For rehabilitation robots to be successful, a control system that results in safe, comfortable and effective interaction between the robot and subject is necessary. We present two types of torque-based controllers for the index finger module of a hand exoskeleton for rehabilitation. Impedance control applies a position-dependent torque at the joints, while keeping the interaction with the finger compliant. Force-field control, on the other hand, applies a position-dependent normal and tangential torques to track a contour in the joint angle space. We also present self-guided impedance control wherein hemiparetic subjects could guide their impaired hand with their healthy hand by using active and passive versions of our device. Experiments with a healthy subject showed that the device could render a range of impedances from low to high. For accurate trajectory tracking high impedance is needed because the inherent finger stiffness varies significantly, however, high impedance increases the torque applied when the finger motion is impeded. On the contrary, force-field control is safer for subjects whose joint show involuntary locking (e.g. spastic joints) and does not apply increased torques during stalled motion, however, provides limited control over the velocity. These results suggest that impedance control would be better suited for therapy where the goal is to train for time-critical tasks and force-field control would work where coordination between the joints is important.