Model Identification and Human-robot Coupling Control of Lower Limb Exoskeleton with Biogeography-based Learning Particle Swarm Optimization

Lower limb exoskeleton is a typical wearable robot to assist human motion and improve physiological power. However, the control performance and stability are affected by some unknown model parameters and control algorithms. Therefore, it is necessary to investigate the model parametric identification and the control design of lower extremity exoskeleton. Firstly, the two degree-of-freedom (DoF) exoskeleton model is constructed by the Lagrange technique. Then the biogeography-based learning particle swarm optimization (BLPSO) is used to optimize the B-spline function parameters and the smooth stimulated trajectories is designed. Meanwhile, the BLPSO is also adopted to identify unknown model parameters of the exoskeleton based on the torques and the joint angles. To decline the negative effect of parametric identification error of exoskeleton, the passive backstepping controller is proposed to improve the tracking performance of human-robot motion. Furthermore, the active admittance controller is adopted to improve the motion comfort of tester. Finally, the comparative experimental results are verified on the platform, which show the BLPSO algorithm has better parametric identification accuracy than PSO and GA. Furthermore, the comparative results have verified that the proposed controller can improve the tracking behavior and reduce the human-robot interaction torque in wearable motion.