Textile Capacitive Touch sensing: Real-Time Localization using Manifold Space Particle Tracking
Textile-based touch sensing offers a novel solution towards creating flexible and robust tactile interfaces that fit within an existing manufacturing ecosystem. Current fabric-based touch sensors can measure distributed and nuanced touch while maintaining textile qualities such as stretching and folding. Recent work as described the development of a single-wire capacitive touch sensing system that consists of a textile interface and external sensing hardware that transmit recorded data to a trained neural network. The sensing method localizes touch along a continuous conductor using a current differential measured at either endpoint. The requisite touch sensing circuit is fully compatible with digital weft knitting, and its layout can be configured in a a variety of pathways to create complete planar textile interfaces. Each interface requires only two connections to external sensing hardware, which allows both the hardware to be reused between interfaces and the data transmitted to the network to have a similar dimension.
Previous work focused on identifying touch location at defined points on a fabric interface through training a recognition system using labeled data. We extend this work towards gesture recognition by using a generated intermediate model to decouple the estimated linear touch location and capacitance from the measured data. The intermediate model forms a manifold of the normalized linear touch location and capacitance within the space of measured data points. A measured data point, acting as a particle, is tracked through the manifold and decoupled to a unique touch location and capacitance. The decoupled linear touch location is mapped to a planar model of the textile interface, which is generated from a user-defined image of the interface. the use of digital modeling tools creates a closed-loop interface design process which allows designers to construct a visual representation of the fabric circuit and predict its sensing performance.
Keywords: Functional Fabrics, e-Textiles, Capacitive Sensing, Signal Processing