Abstract
This paper proposes a novel control algorithm for torque-controlled exoskeletons assisting cyclic movements. The control strategy is based on the injection of energy parcels into the human-robot system with a timing that minimizes perturbations, i.e., when the angular momentum is maximum. Electromyographic activity of main flexor-extensor knee muscles showed that the proposed controller mostly favors extensor muscles during extension, with a statistically significant reduction in muscular activity in the range of 10–20% in 60 out of 72 trials (i.e., 83%), while no effect related to swinging speed was recorded (speed variation was lower than 10% in 92% of the trials). In the remaining cases muscular activity increment, when statistically significant, was less than 10%. These results showed that the proposed algorithm reduced muscular effort during the most energetically demanding part of the movement (the extension of the knee against gravity) without perturbing the spatio-temporal characteristics of the task and making it particularly suitable for application in exoskeleton-assisted cyclic motions.
Highlights
To validate the effectiveness of the controller, Active Robot (AR) mode and Zero Torque (ZT) mode were compared in terms of knee joint angle q, angular velocity qand muscular activity of three extensor muscles (RF, Vastus Medialis (VM), Vastus Lateralis (VL)) and two flexor muscles (BF, ST)
It is worth noticing that during the oscillation cycle a single energy parcel provided an amount of energy with respect to the total energy so that ǫ = 19% on average, demonstrating that the controller did account for the total energy cost of the task with a non negligible contribution
The controller was tested on a group of healthy subjects performing knee flexion-extension assisted by a compliant exoskeleton
Summary
Wearable exoskeletons are being developed for the therapy of patients undergoing neuro-motor recovery, for the daily-life support of subjects with permanent motor impairments and for the assistance of healthy individuals in industrial applications.Human-robot mutual adaptation is a key mechanism to be considered in the control of wearable exoskeletons: robots have to synchronously adapt to the intended motion of the user, who in turn should be allowed to exploit robotic physical support to improve his motor function, in case of an impaired subject, or to possibly reduce the effort to perform a task, in case of a healthy subject.These aspects strongly motivate the need to detect user’s motion intention and to adjust robot action in a smooth, natural and non-constraining way.Many approaches have been explored to assist human movements compliantly and adaptively, including interfaces for intention recognition based on biosignals (electroneurographic or electromyographic measurements) or on motion reconstruction/prediction (kinematic or dynamic measurements) (Tucker et al, 2015; Yan et al, 2015).The first approach includes solutions that might be invasive, unreliable or sensitive to calibration, repeatability and signal acquisition/processing issues. Human-robot mutual adaptation is a key mechanism to be considered in the control of wearable exoskeletons: robots have to synchronously adapt to the intended motion of the user, who in turn should be allowed to exploit robotic physical support to improve his motor function, in case of an impaired subject, or to possibly reduce the effort to perform a task, in case of a healthy subject. These aspects strongly motivate the need to detect user’s motion intention and to adjust robot action in a smooth, natural and non-constraining way.
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
