Adaptive Oscillator-Based Gait Event Detection and Trajectory Planning for Lower Limb Exoskeleton

Lower limb exoskeletons have been shown to have the potential to assist in patient rehabilitation and improve walking ability. How to accurately determine the patient's movement intention and plan an appropriate trajectory has been an important challenge in the control of lower limb exoskeletons. In this paper, we proposed a gait trajectory planning algorithm for lower limb exoskeleton based on adaptive oscillator and deep neural network to achieve continuous motion estimation of lower limb joints, which can adapt to changes in different walking speeds and maintain the desired trajectory. The proposed algorithm used real-time joint angle data to classify gait phases using neural network algorithms to determine gait characteristic event points. Then, we built an adaptive oscillator model to estimate the gait phases of the motion in real time and realize the online planning of the gait trajectory. A preliminary experimental study was conducted to validate the proposed online gait trajectory planning algorithm. The experimental results show that the online planning algorithm can adapt to different walking speed environments and obtain suitable desired trajectories for the exoskeleton.