Research ArticleAdaptive control of 5 DOF upper-limb exoskeleton robot with improved safety

•The proposed reliable controller design considers the modeling errors and the uncertainties to improve tracking performance.•The proposed controller can guarantee the safety of the patients against large parameter variance, hardware malfunctions, even actuator faults.•The estimation of velocity and acceleration provided by an adaptive observer without adding more sensors.•The effect of the hardware malfunctions, even actuator faults is described in a unified model and automatically accommodated.•The switching logic in the controller and the gain updated law will provide a natural and effective fault accommodation mechanism.

This paper studies an adaptive control strategy for a class of 5 DOF upper-limb exoskeleton robot with a special safety consideration. The safety requirement plays a critical role in the clinical treatment when assisting patients with shoulder, elbow and wrist joint movements. With the objective of assuring the tracking performance of the pre-specified operations, the proposed adaptive controller is firstly designed to be robust to the model uncertainties. To further improve the safety and fault-tolerance in the presence of unknown large parameter variances or even actuator faults, the adaptive controller is on-line updated according to the information provided by an adaptive observer without additional sensors. An output tracking performance is well achieved with a tunable error bound. The experimental example also verifies the effectiveness of the proposed control scheme.