Disturbance-Observer-Based Sliding Mode Controller for Physical Human–Robot Interaction in Multi-Joint Exoskeleton Robots

Abstract

A robust and stable physical Human-Robot Interaction (pHRI) controller is crucial for robotic systems, particularly for rehabilitation devices that enable continuous human-robot interaction. A key challenge is to achieve a precise interaction torque that enables the robot to provide active assistance or remain transparent to the user. This paper proposes a novel direct force controller, termed as Integral Sliding Mode Controller with a Disturbance Observer Based on the Interaction torque model (ISMCDOBI), for pHRI control of multi-joint exoskeleton robots. A spring-based contact model is explicitly integrated into the robot dynamics to account for human–robot interaction effects and to compensate for human motion. Unlike SEA-based or position/velocity-based methods that indirectly regulate interaction forces, the proposed approach operates directly in the force/torque domain. An integral sliding mode controller is combined with a disturbance observer that employs a nominal interaction model to account for uncertainties and disturbances. Gravity, friction, and human motion compensators are also incorporated to reduce control effort and improve the tracking of the desired interaction torque. The effectiveness of the proposed framework is validated through realistic MATLAB–OpenSim co-simulation on a coupled human–exoskeleton biomechanical model. The results demonstrate the effectiveness of the proposed interaction controller by maintaining interaction torques less than 0.42 N·m and 0.52 N·m in transparent mode for the hip and knee. Also, a one-way ANOVA analysis depicts the superiority of the proposed controller over two common torque control methods in various modes.