Loopsense: minimal-sized, multi-DOF knitted force sensors (2024)
credit: Media Interaction Lab
Knitted resistive sensors are commonly very limited in terms of design choices and sensing versatility, since they take up a relatively large area and considerably limit the choice of knitting structure. As a sub-project of TextileUX, the Media Interaction Lab investigated knitting of wires with piezoresistive coating to create resistive force sensor cells. The resulting research paper Loopsense presents those sensors, which can consist of only a single pair of loops in their most small-scale implementation. As they only affect a low number of stitches, they can be easily integrated into a variety of flatbed knitted fabrics and – depending on the nature of the pattern – they may be easily concealed from the fabric's surface. This provides great latitude for textile designers, enabling them to access a vast range of possible knitting patterns. Beyond visual aesthetics, this has considerable implications for haptic and physical design choices, since a knitting pattern also affects properties like stability, elasticity, anisotropic strain behavior, etc.
Another limitation of traditional knitted force sensors is that the actuation modality is usually unknown, unless additional state information is available, which is rarely the case. This means for example that, since every application of force modifies the sensor's electrical resistance, strain and pressure cannot be discerned, which can easily lead to false positives and/or distorted measurement data. Utilizing a variety of different sensor implementations by modifying the local loop meshing, we are able to create resistive sensor cells that largely vary in their response and characteristics, with respect to actuation modality. By harnessing the variety provided by flatbed and v-bed knitting, we combined different stitch types to establish sensors that vary in their response to certain actuation modes. Selectively combining those, we are able to infer multiple degrees of freedom (DOF) from the input signals and, e.g., tell in what direction the textile device was strained or if it was pressed.
In our awarded (best paper honourable mention) ACM CHI 2024 paper, we present in-depth information about fabrication, loop meshing, structural integration, and a prototypical sensor fusion approach to differentiate between orthogonals strain direction and surface pressure. We furthermore show a line of three exemplary two-faced fabrics concealing the sensors from the front face, as well as use case scenarios, including a knitted glove that measures finger bend, a 2D strain-based input device, and a minimalist musical keyboard that supports key velocity and pitch bend. For more details please see the paper (Open Access). In the paper's supplementary material, we provide more details about implementation of the data processing pipeline, additional sensor metrics, Knitout based knitting description files, as well as firmware code used for sensor sampling.
Credits: University of Applied Sciences Upper Austria
credit: Media Interaction Lab
