A FORCE MYOGRAPHY-BASED SYSTEM FOR GAIT EVENT DETECTION IN OVERGROUND AND RAMP WALKING

Abstract

In this paper, we present a novel method to determine the heel strike (HS) and toe-off (TO) during overground (OG) and ramp walking, including the transition. The method utilizes force myography (FMG) signals from thighs while subjects walked on OG and ramp. Five adult male subjects wore a wireless FMG data acquisition system, developed in-house using force resistive sensors and electronic components. A heuristic approach for subject-dependent and terrain-independent model was developed to determine HS and TO in a given gait cycle in steady state and transition. The average error in HS determination was 9.66 ± 8.29, 9.38 ± 9.35, and 13.94 ± 18.95 ms, while TO was determined with an average error of 16.99 ± 18.12, 13.35 ± 15.10, and 17.29 ± 21.92 ms for OG, ramp, and transition, respectively. The proposed system is less expensive, simple to develop, and friendly to wear. The reported errors are comparable to previously reported errors using pressure sensitive insole, gyroscope, accelerometers, and electromyography, which are much complex and expensive in comparison to proposed FMG-based system. Although the tests were conducted on healthy subjects, the system promises to be generalizable to amputee and other pathological gaits also. While the tests were conducted on young adults at self-selected speeds, the system also promises to be generalizable for a wide range of walking speeds across the population.

Existing System

Heel strike (HS) and toe-off (TO) method, force myography (FMG) method.

Proposed System

In this paper, a new method has been presented to detect the HS and the TO using an FMG-based system during the OG walk, ramp walk, and transition between them. The proposed FMG signal acquisition system with the proposed algorithm was able to estimate all HSs and TOs from a full length trial with errors comparable to the existing systems. The system can be mounted or fixed to the thigh with less obstruction and higher mobility, which has always been a problem with inertial measurement unit-based systems. The placement of sensor seems to affect the performances of the proposed measuring system as R8 could perform better than other FMG sensors. Signals from FMG sensors are first preprocessed and fed to the signature extraction block which extracts one signature for every sample of the incoming signal. During the training phase, the signatures corresponding to true events are stacked together to form a signature database. During the testing phase, the signature for every sample is matched with the signatures stored in the signature database. Finally, the events are detected by detecting peaks in the outcome of the signature matching stage.

Architecture

BLOCK DIAGRAM