An Adaptive Oscillators-Based Approach to Achieve Transparent Control of a Six DoF Lower-Limb Exoskeleton

Abstract

Lower-limb exoskeletons have been effectively used in rehabilitation to reduce gait impairments. However, making the exoskeleton transparent to provide suitable net interaction torques and assist patient’s movements is yet an open challenge. In this study, we designed a torque control method based on adaptive oscillators (AOs) to reduce interaction forces with a six degree-of-freedom exoskeleton, named the ExoRoboWalker. A synchronization layer was designed with a pool of AOs to estimate user’s gait phase, and a baseline torque controller was introduced to generate a torque profile based on user-robot interaction during the previous gait cycles. A zero-torque controller was used in parallel to the main torque controller with the objective of improving users’ control of the system and their balance. The proposed controller was experimentally evaluated in six healthy subjects who walked with the exoskeleton for 200 gait strides at different walking speeds. A transparent controller we previously implemented based on a zero-impedance model was also evaluated and results were compared with the new controller. The controller based on AOs was able to reduce the average interaction torques by 40% at the highest gait speed of 0.8 m/s relative to the zero-impedance controller we previously implemented.