Synergy-Based Control of Assistive Lower-Limb Exoskeletons by Skill Transfer

Abstract

Considering neuronal coordination between limbs, this article presents a study on the control of lower-limb exoskeletons for assistance of human gait by transferring the motor skills. The synergy-based robotic controller captures kinesiological information and biological signals from the healthy leg and generates intended motor patterns for the assisted leg in different gait phases of the slope walking behavior. First, we have developed a computationally efficient stiffness estimation model of the lower-limb joints and identified the experimental parameters in accord with the subject's locomotion behavior. The estimated stiffness matrix at minimum muscular contraction is scaled by cocontraction index and mapped to joint stiffness to be utilized in the control design. Then, we have proposed the impedance matching model and realized human skills transfer by surface electromyography signals. Considering the uncertain dynamics of the human-robot system, we have developed an adaptive fuzzy approximator to estimate robot's dynamic parameters and drive the robot tracking the referenced trajectories. The developed synergy-based control has been verified using three subjects with varying locomotor abilities. Results from these participants have shown a symmetrical and consistent adaptability between two legs with the synergy-based control, while the range of motion of the assisted leg in the affected side is more volitional and individualized.