Human-Robot Modeling of the Maestro Hand Exoskeleton

Based on previous works by Agarwal and Esmatloo, I re-derived the forward kinematics and inverse kinetics of the human-robot exoskeleton system and implemented it in real-time within Maestro’s C++ libraries.

The Maestro hand exoskeleton operates at three coordinate domains: robot joint space, finger joint space, and fingertip Cartesian space. Based on previous work and four-bar linkage principles, I calculated the human-robot system’s forward kinematics and inverse kinetics. The forward kinematics use robot joint angle measurements to estimate fingertip position, which can then be used to implement virtual environments at the fingertip. The inverse kinetics are used to map desired fingertip interaction forces back to desired robot joint torques, which are then commanded to the Maestro hand exoskeleton. This project highlighted importance considerations and shortfalls of the modeling process, such as sensitivity to parameter errors and sensor calibration.