Simulation study on the design and control of humanoid shoulder joint based on pneumatic artificial muscle drive

This paper presents a simulation study of a three-degree-of-freedom (3-DoF) humanoid shoulder joint system driven by pneumatic artificial muscles (PAMs). A 6-PAM-based joint mechanism is designed to achieve compound motions, including flexion–extension, internal–external rotation, and abduction–adduction via a rotary antagonistic structure. The mechanical characteristics of the PAMs are identified using a particle swarm optimization (PSO) algorithm, and a nonlinear model incorporating hysteresis and coupling effects is established. To enhance control performance, a sliding mode controller (ASTSMC) based on a saturated sliding surface and adaptive super-twisting disturbance compensation is proposed. The global asymptotic stability of the closed-loop system is rigorously proven using Lyapunov theory. Simulation experiments are conducted in MATLAB under three scenarios: step response, sinusoidal tracking, and disturbance rejection. Results show that the proposed controller reduces the average steady-state error (SSE) to 0.00027[Formula: see text]rad with a maximum overshoot of only 0.22% in the step response test. For sinusoidal tracking, the average root-mean-square error (RMSE) is 0.00020[Formula: see text]rad. In the disturbance rejection test, the average SSE is suppressed to 0.00155[Formula: see text]rad. These results verify the proposed controller’s superior response speed, accuracy, and robustness under dynamic loading conditions.