Fully textile passive wireless sensing for human movement monitoring with multiple sensors

TL;DR

This paper introduces a fully textile wireless system that can monitor human movement using multiple sensors, offering comfort and real-time tracking without rigid components.

Contribution

The novelty lies in using a single inductor for multiple capacitive strain sensors in a passive wireless system for movement monitoring.

Findings

The system achieved an F1-score of 0.98 for static and 0.96 for dynamic activities with two integrated sensors.

Accuracy and F1-score dropped to 0.86 and 0.87 when testing across three independent sessions.

The system demonstrates potential for unobtrusive smart clothing for real-time movement monitoring.

Abstract

Movement monitoring with wearable technologies is becoming increasingly popular in different fields of application (clinical, sports, entertainment). Particularly, textile-based wearables for movement monitoring are attractive as they follow the body movement, are comfortable to use, and can provide continuous tracking capabilities. Ideally, these wearable devices should be flexible (as opposed to current technologies with rigid electronics on the garments) and transmit data wirelessly to avoid hindering the natural movement with connections. Although fully textile wireless and passive wearable systems — whereby the textile sensing part does not have any rigid components and the data is wirelessly transmitted to an external reader — have been developed, the capability of these technologies is currently limited to a single sensor. In this work, we present a system based on a resonating inductor-capacitor (LC) circuits that can measure multiple sensors to broaden the range of use by tracking more than a single joint. Importantly, the presented system employs multiple capacitive strain sensors but retains the use of a single inductor for data transmission, limiting the complexity of realization and the number of connections. After characterization on the bench for careful design of the circuit components, we demonstrated the capability of the system to be used for human movement monitoring and activity classification by integrating two sensors in sport leggings and performing different static and dynamic activities. The tests with sensorized leggings were performed by a single participant. Among a set of chosen classification algorithms, the best performance (F1-score) was 0.98 for the static activities and 0.96 for dynamic activities. When including three independent sessions (donning and doffing the sensorised leggings) accuracy and F1-score dropped to 0.86 and 0.87 respectively. Overall, the presented system has the potential to be adopted as unobtrusive and comfortable smart clothing for real time movement monitoring.

Figures