Gait trajectory prediction for lower-limb exoskeleton based on Deep Spatial-Temporal Model (DSTM)

Gait trajectory prediction can not only be used to reduce the interactive resistance for passive powered exoskeleton (wearer leading exoskeleton to walk), but also be used to generate suitable gait trajectory for active powered exoskeleton. This paper proposes a gait trajectory prediction methodology using Deep Spatial-Temporal Model to predict the lower-limb joint trajectories of future gait trajectory according to historic gait data. The method can obtain the deep spatial-temporal relationship of gait trajectories by training large gait data, which avoids the complex modeling of human body and lower-limb exoskeleton. In order to make the prediction results more stable, two-step predictions are adopted to estimate the optimal future gait trajectory. Based on a gait set including 35 subjects, the experiments demonstrate that the gait trajectory prediction methodology can accurately predict the future gait trajectory.