In order to develop the application of pneumatic muscle in multi-freedom rehabilitation robot, a 2-DOF lower limb rehabilitation auxiliary training robot was designed for gait rehabilitation training. The actuator was driven by pneumatic muscle and stretch spring in parallel to realize energy-saving assisted walking.The kinematics and dynamics models of human lower limbs were established, and the standard CGA gait curve was used as the joint input, and the kinematics simulation was carried out by SolidWorks/Motion, which verified that the designed model was consistent with the motion law of human lower limbs. Aiming at the difficulty of joint control due to nonlinearity in pneumatic muscle extension, a fuzzy adaptive PID control algorithm was proposed. MATLAB/Simulink was used to perform simulation control experiments. The simulation results show that the fuzzy PID adaptive control algorithm can achieve better following effect than the traditional PID. Finally, through the experimental platform, it is verified that the fuzzy PID adaptive control algorithm can meet the needs of patients active rehabilitation training.