Adaptive Suppression of Severe Pathological Tremor by Torque Estimation Method

Abstract

Tremor is a rhythmical and involuntary oscillatory movement of a body part. In addition to social embarrassment, tremor can be debilitating for daily activities. Recently, wearable active exoskeletons emerged as a noninvasive tremor suppression alternative to medication or surgery. The challenge in musculoskeletal tremor suppression is identifying and attenuating the tremor motion without adding resistance to the patient's intentional motion. In this research, an adaptive tremor suppression algorithm was designed to estimate the tremor fundamental frequency and calculate the proper suppressive force to be applied by the orthosis to the patient's arm. Stability of the closed-loop system and robustness against the parametric uncertainties were analyzed. An experimental setup was designed and developed to emulate the dynamics of a human wrist with intentional and tremor motion. A pneumatic cylinder and a sliding mode integral controller was used to apply orthotic suppressive force. The algorithm was implemented with an NI cRIO real-time controller and tested using clinical data from ten patients with severe pathological tremor. Experimental results showed tracking of the tremor frequency with less than 3-s response time, and an average 34.5 dB (98.1%) and 11.8 dB (74.3%) reduction of tremor amplitude at the fundamental and second-harmonic frequencies, respectively. The average resistance force to the intentional motion was 0.7 N and the average position error was 2.08% (0.18 dB). The results were compared with passive tremor suppression using a tunable magnetorheological damper.