To solve the problems of external interference and system parameter uncertainty in the trajectory tracking of a lower limb rehabilitation robot driven by pneumatic artificial muscles, an active disturbance rejection algorithm for joint control is proposed. Based on the mathematical model of the servo control system of the pneumatic artificial muscle joint, the method firstly estimates the system state and disturbance using a third-order state expansion observer. It then compensates for the disturbance in real time and adjusts the parameters based on the separation principle. Subsequently, with using a pneumatic artificial muscle test platform, the step signal tracking control, square wave tracking control, and sine tracking control of the control system are compared and verified under fixed angle conditions. Experimental results show that the designed active disturbance rejection control (ADRC) has a faster response time and lower control error than the proportional integral differential (PID)controller, meeting the application control requirements of the lower limb rehabilitation robot.