A Model-Based Method for Minimizing Reflected Motor Inertia in Off-board Actuation Systems: Applications in Exoskeleton Design

The research and development of wearable robotic devices has been accelerated by off-board control and actuation systems. While off-board robotic actuation systems provide many benefits, the impedance at the robotic joint is often high. High joint impedance is undesirable for wearable devices like exoskeletons, as the user is unable to move their joint without actively controlled motion from the motors. We propose that the impedance can be reduced substantially in off-board robotic actuation systems by minimizing the reflected inertia from the motor. We have developed a model and optimization-based methodology for selecting a motor and set of mechanical design parameters that minimize reflected inertia. This methodology was implemented in the design of an off-board knee exoskeleton as a case study. A grey-box model was developed that incorporates biomechanical knee trajectories, an experimentally determined human-device interface stiffness model, Bowden cable stiffness and friction, and a motor model. A constrained optimization routine was developed that uses the model and a library of157 candidate servo motors to select the actuator and mechanical design parameters that minimize reflected inertia at the exoskeleton joint. We found that S6 of the motors were able to carry out the necessary torque-velocity trajectories to achieve the prescribed exoskeleton joint torques and limb motions. The optimal motor was the Kollmorgen C133A-one of the largest in the library of candidate servo motors and required a 2.25 cm actuator pulley at the knee joint and a 17.5 cm cable sheave at the motor output. This methodology can be adapted by exoskeleton designers to develop more backdriveable exoskeletons and improve experimental capabilities. All code developed for the case study is open-source and freely available online.