U2FSM: Unsupervised Square Finite State Machine for Gait Events Estimation from Instrumented Insoles

Luigi D’Arco, Haiying Wang, Huiru Zheng

Research output: Chapter in Book/Report/Conference proceedingChapterpeer-review

Abstract

Gait analysis is a research field that aims to assess and analyse a person’s locomotion patterns. Traditional methods rely on visual evaluations by medical experts, but recent advances in biomechanics have introduced objective solutions such as motion capture systems, force plates, and pressure mats. However, these solutions are expensive, cumbersome, and limited to controlled environments. This paper proposes a novel hybrid model for gait event detection using instrumented insoles with pressure sensors. The model combines a finite state machine and a fuzzy c-mean algorithm to accurately identify gait events, including heel strike, foot flat, heel off, and toe-off. Random sampling was employed to evaluate the model’s performance, ensuring representative results across the population. Nine parameters, including the duration of events, stride, stance and swing duration, and percentage of stance and swing phases were the main focus of the evaluation. The proposed system demonstrated accurate recognition of step counts and duration, with minimal variations compared to manually annotated data. Although a 0.1 s overall error in the duration of the gait events was identified, favouring longer heel strikes over shorter foot flat events, this was attributed to the amplitude-based annotation process’s constraints. The proposed solution aligned with the optimal percentages of the stance and swing phases according to the gait cycle model. The results indicate the reliability and potential applicability of the proposed system in real-world scenarios. Future research will focus on refining the model, addressing observed errors, and exploring additional gait parameters to provide a comprehensive analysis of human locomotion patterns.
Original languageEnglish
Title of host publicationAdvances in Computational Intelligence Systems
PublisherSPRINGER LINK
Pages273-285
Number of pages13
ISBN (Print)9783031475078, 9783031475085
DOIs
Publication statusPublished online - 1 Feb 2024

Fingerprint

Dive into the research topics of 'U2FSM: Unsupervised Square Finite State Machine for Gait Events Estimation from Instrumented Insoles'. Together they form a unique fingerprint.

Cite this