Real-Time Gait Phase Detection Based on LSTM-ResMLP-LightGBM Approach for Exoskeleton in Outdoor Activity

Real-Time Gait Phase Detection Based on LSTM-ResMLP-LightGBM Approach for Exoskeleton in Outdoor Activity

Blog Article

Gait phase detection plays a significant role in human motion intention recognition in lower limb exoskeleton.This study proposes a LSTM-ResMLP-LightGBM method for real-time gait phase detection in multigrade loads in outdoor activity.LSTM is exploited to extract the underlying meaning and construct a lengthy time relationship in multiple features.

ResMLP effectively extracts key information to minimize multi-feature redundancy.LightGBM is an external framework based on gradient boosting decision tree with leaf-wise growth strategy.There bosch 4100 table saw motor replacement are five locomotion modes in diverse terrains, such as LW, SA, SD, RA, and RD.

In every locomotion mode, a gait cycle is divided into five subphases, such as HS, MS, HO, PW, and AS.The time delay for actual gait phase detected by the hybrid method is distributed in multigrade loads in five locomotion modes.The proposed method has a prominent effect on gait phase detection in terms samsung a71 price toronto of assessment criterion of MSE, MAE, time delay, accuracy, precision, recall, F1 score, and confusion matrix.

What’s more, compared with the general approaches of TST, TFT, TCL, and NGBoost, the proposed method makes great influence on gait phase detection in five locomotion modes.In addition, the mean early detection time is 63.79 ms that contributes to enhance the response effect of lower limb exoskeleton.

Finally, this study supplies a feasible and stable method of gait phase detection for human-robot cooperative walking.