Characterization and Control of a Woven Biomimetic Actuator for Wearable Neurorehabilitative Devices

Strokes are a major cause of death and disability, with nearly 300,000 Canadians suffering from stroke-related disability. Many survivors suffer from upper limb paresis, which makes ordinary tasks difficult. Availability of care is an issue, with only 43.3% of patients in Ontario receiving proper treatment. Robot-assisted neurorehabilitation has been used to address limited clinician availability, and has been shown to improve patient strength and range of motion faster than or comparable to traditional methods. However, the size and weight of current devices limit their use to within therapists’ offices. Incorporating actuators into clothing would make the devices more comfortable, natural to use, and allow for at-home care, which has been shown to improve therapy compliance, frequency, and efficacy. In this work, a woven fabric bi-pennate artificial muscle system is presented. Its contractive stroke, force, speed, and efficiency were characterized, and the transient stroke and force were modelled. The latter was used to create a composite feedforward-feedback controller, which was tested to determine its performance in controlling the actuator using solely force feedback. The maximum stroke, force, and controlled contraction period were found to be 12.23%, 11.28 N, and 2.60 s respectively, with the first two scaling linearly with the number of active actuators. Results indicate that the actuator is able to meet or exceed clinical performance requirements, albeit nonconcurrently, and that the designed controller was an effective force feedback control method, limiting overshoot to 2.6% and steady state error to 0.94%. Further work is suggested to refine the actuator’s design.