Towards the Development and Characterization of a Torque Sensor for Volitional Control of a Pediatric Exoskeleton

Each year in the U.S., between 1200 and 1500 school-aged children are diagnosed with cerebral palsy, a neurological disorder affecting body movement and the nervous system. The development of lightweight and custom exoskeletons to assist users during ambulation has experienced a fast growth in the last decade as a strategy to assist children with locomotion disabilities. Measurement of the torque exerted at the joints of the exoskeleton is crucial to allow volitional control from the users and maximize therapeutic outcomes, by providing information about user and robot interaction. Commercial torque sensors are characterized for being too heavy and bulky to be implemented for pediatric exoskeleton applications. In this study, we have developed a custom torque sensor utilizing shear strain gauges to estimate joint torques in an exoskeleton. The strain gauges were mounted onto the output plate of the exoskeleton joint forming 2 Wheatstone Bridges at 90-degree angles apart from each other, and the signal was amplified and digitized for better signal resolution. Static tests were conducted by applying known weights to the system and comparing measured output values to a reference torque transducer. The system was tested at a maximum torque of 85 [Nm], and initial results show a nonlinearity of less than 15 percent of full scale, suggesting that the proposed strain gauge system is valid for our application. Future work includes performing static tests at other angles, and performing dynamic tests that resemble lower limb joint angle trajectories in human walking.