Research on human gait prediction and recognition algorithm of lower limb-assisted exoskeleton robot

Wearable lower limb-assisted exoskeleton robot can improve a human's ability to walk long distances under heavy load. Accurate perception and recognition of human lower limb motion and real-time input of exoskeleton control system parameters have always been the key technologies of wearing lower limb assist exoskeleton robot. To solve the problem of poor real-time and stability of the lower extremity exoskeleton system caused by time delay in transmitting gait information from the lower extremity assisted exoskeleton robot perception system to the control system, we proposed a new human gait prediction and recognition algorithm. Based on the gait sample data of human lower limbs, we proposed a prediction algorithm concerning least-squares support vector regression (LS-SVR) to predict the gait data at the next moment of human lower limbs movement. We proposed a human gait phase recognition algorithm using fuzzy theory to identify the gait phase at the next moment of human lower limb movement. The proposed algorithm can identify the heel landing phase, the load-supporting phase, the pre-swing phase, and the swing phase of human lower limbs. Experimental results demonstrate that gait prediction accuracy is 99.83% and the gait phase recognition rate based on the predicted gait data of 120 complete gait cycles is 99.93%. It comes up to a feasible method for wearable lower limb-assisted exoskeleton robot intelligent perception technology. New algorithms improve the recognition rate of gait phase prediction and effectively improve the real-time, stability, and robustness of the lower extremity exoskeleton robot system. The proposed gait prediction and recognition model can help gait recognition tasks to overcome the difficulties in real application scenarios and provide a feasible method for a lower extremity assisted exoskeleton robot perception system.