An exciting new direction in ubiquitous computing is the integration of textile-based sensing elements directly into everyday clothing. However, the design of smart textiles that weave together these elements presents several system design challenges, including sensor, hardware, algorithm, and application aspects. Our project tackled these challenges and delved into several important verticals. 

We first explored the design of comfortable, loose-fitting smart garments. We questioned whether familiar garments made from materials like cotton and silk could be subtly adapted for sensing human activity and physiology. Our goal was to yield naturally fitting, comfortable, and less obtrusive smart clothing.

We recognized that although numerous textile-based pressure sensors existed, their practicality was limited by narrow detection ranges, an inability to simultaneously sense static and dynamic pressures, and low durability. Thus, one of our core challenges was to design a humidity-resistant, all-fabric pressure sensor with high sensitivity across a broad range of pressures—from subtle heart pulses to body posture—exceeding the capabilities of previously reported sensors.

This sensor was leveraged in several novel applications. The first, Phyjama, is a comfortable sleep garment capable of measuring heart ballistics signals. We integrated our developed pressure sensors into loosely-worn sleepwear to measure physiological signals. Our design uniquely combines textile elements in sensed regions with discrete electronic components only in specific areas, such as buttons. It fuses signals from a distributed set of sensors, enabling the detection of heart rate, breathing rate and body posture signals and works with loose-fitting clothing by opportunistically leveraging any form of contact between loose fabric and the body. This work was published in ACM IMWUT 2019.

Another innovative application was the design of interactive soft toys embedded with an array of new sensors for fine-grained interaction detection. These “FabToys” are equipped with a 24-sensor array of fabric-based pressure sensors beneath the surface, ensuring dense spatial sensing coverage while maintaining the natural feel and softness of the fabric. We optimized both the hardware and software pipeline, achieving high accuracy in detecting a wide range of interactions in different regions of the toy. Our contributions included sensor array fabrication, data acquisition and triggering methods, and neural network models with early exit strategies for computational efficiency and autoencoder-based channel aggregation for communication optimization. This work was published at ACM MobiSys 2022 and was selected as an ACM SIGMOBILE Research Highlight.

The second direction we pursued was the design of new comfortable EEG electrodes and enhanced textile-based sleep monitoring garments.

We designed a new thread-based, reusable wet electrode that addressed the fact that dry electrodes in wearable devices had significant motion artifacts, whereas wet electrodes lacked aesthetic and comfort. Our electrode combines the high signal quality of commercial wet electrodes with the comfort and unobtrusiveness of dry electrodes.

In Matter 2020, we described a textile-based system that enhances a popular sleep mask with this new electrode. Recording these signals near the face is challenging, as most subjects are sensitive to implements placed on the face or head. Our design synergistically uses fabric electrodes and garment strategies to create an unobtrusive platform. We extended this work to develop PhyMask, a fully validated clinical-grade wearable sleep monitoring system. Unlike traditional wearables, PhyMask comfortably acquires all signals relevant to sleep solely using textile sensors placed on the head. PhyMask accurately measures all signals required for precise sleep stage tracking and robustly extracts advanced sleep markers such as spindles and K-complexes in real-world settings. Validated against polysomnography, PhyMask significantly outperforms commercially available sleep tracking wearables like Fitbit and Oura Ring. This work was published at ACM Health 2022.

The third direction we explored was addressing communication challenges with antenna arrays on textile-based substrates. Embedding antenna arrays in textiles like curtains can enhance through-wall sensing and beamforming for IoT devices, and improve indoor localization of Bluetooth tags. 

However, deformation of textile-based antennas alters their behavior. We developed CurtainNet, a flexible UHF-band antenna array on a large curtain surface, using a combination of optical and RF tracking to compensate for deformations, occlusions, and phase changes. CurtainNet significantly outperforms alternative methods in beamforming performance and increases indoor range. This work was presented at ACM SenSys 2023.

Collectively, the work completed in this project explores new avenues in textile-based sensing, wearable system design, and the practical applications of this innovative technology.

Last Modified: 11/21/2023
Modified by: Deepak K Ganesan