A Robust Wheel Interface with a Novel Adaptive Controller for Computer/Robot-Assisted Motivating Rehabilitation

Document Type

Conference Proceeding

Publication Date


Source ISSN



TheraDrive is an effective system for post-stroke upper extremity rehabilitation. This system uses off-the-shelf computer gaming wheels with force feedback to help reduce motor impairment and improve function in the arms of stroke survivors. Preliminary results show that the TheraDrive system lacks a robust mechanical linkage that can withstand the large forces exerted by patients, and it lacks a patient-specific adaptive controller to deliver personalized therapy. It is also not capable of delivering effective therapy to severely low-functioning patients. A new low-cost, high-force haptic robot with a single degree of freedom has been developed to address these concerns. The resulting TheraDrive consists of an actuated hand crank with a compliant transmission. Actuation is provided by a brushed DC motor, geared to output up to 23 kgf at the end effector. To enable a human to interact with this system safely, a special compliant element was developed to double as a failsafe torque limiter. A set of strain gauges in the handle of the crank are used to determine the interaction forces between human and robot for use by the robot’s impedance controller. The impedance controller is used to render a one-dimensional force field that attracts or repels the end effector from a moving target point that the human must track during therapy exercises. As exercises are performed, an adaptive controller monitors patient performance and adjusts the force field accordingly. This allows the robot to compensate for gravity, variable mechanical advantage, limited range of motion, and other factors. More importantly, the adaptive controller ensures that exercises are difficult but doable, which is important for maintaining patient motivation. Experiments with a computer model of human and robot show the adaptive controller’s ability to maintain difficulty of exercises after a period of initial calibration.


Published as part of the proceedings of the conference, ASME/ISCIE 2012 International Symposium on Flexible Automation, 2012: 87-94. DOI.