Human Gait phases recognition based on multi-source data fusion and BILSTM attention neural network

• Published in: Measurement : journal of the International Measurement Confederation Vol. 238; p. 115396
• Main Authors: Zhan, Haoran, Kou, Jiange, Cao, Yuanchao, Guo, Qing, Zhang, Jiyu, Shi, Yan
• Format: Journal Article
• Language: English
• Published: Elsevier Ltd 01.10.2024
• Subjects: BILSTM attention neural network; Gait recognition; Lightweight gait acquisition device; Lower limb exoskeleton; Multi-source data fusion; Lower limb exoskeleton; Lightweight gait acquisition device; Gait recognition; BILSTM attention neural network; Multi-source data fusion
• ISSN: 0263-2241
• DOI: 10.1016/j.measurement.2024.115396

A human gait recognition algorithm with deep learning based on multi-source data fusion is proposed to assist exoskeleton realizing complex human-exoskeleton cooperative motion. A lightweight gait acquisition device simultaneously acquires three different types of sensor signals, i.e., thigh surface electromyography (sEMG), ground reaction force (GRF) of human feet and two joints motion information. The sample data is obtained by many multi-scene gait experiments involving stand still, walk up/down stairs, and walk up/down slope, etc. The proposed algorithm recognizes multiple gait phases (27 classes) in different motion patterns by using short-time gait data, which has two typical advantages: (1) Based on multi-source sensor distributed in human body, this wearable device is constructed for ease of use to address several Subjects (different weights and heights) with low cost and light weight. (2) The critical features of gait data are identified by Bi-directional Long Short-Term Memory (BILSTM) attention neural network with higher recognition accuracy than other state-of-art recognition methods, especially in arbitrary gait switching durations. [Display omitted] •A new portable gait acquisition device is designed to simultaneously acquire three different types of sensing signals, including human sEMG, lower limb GRF, and knee/hip joint motion, as shown in Fig.1. This wearable device is constructed for ease of use to address several Subjects (different weights and heights) with low cost and light weight.•A multi-source data fusion and BILSTM attention neural network is proposed to identify detailed gait phases in multiple motion patterns with high accuracy.•Consider two indices such that the false detection number and rate are used to verify the effectiveness of the gait recognition in arbitrarily gait switch durations. Furthermore, two different subjects data are used to verify the effectiveness of the proposed algorithm comparing with the other two state-of-art algorithms.