We demonstrate how to reduce sensor data inconsistencies of knitted strain sensors, using a low-scale, computationally inexpensive, Artificial Neural Network. As features, we use raw sensor data that is temporally smoothed using a set of different smoothing factors.

We systematically optimized the performance of Twill-based knitted strain sensors by studying optimizing knitting structure and material composition. We modified stitch settings, thread count, and knit meshing. Furthermore, we improvement elastic recoil and reduced wear-out effects by adding Lycra to the substrate yarn.

We present a design for knitted tactile UI elements, that are soft, compressible, and stretchable. They sense pressure with high sensitivity and responsiveness. Using a regular flatbad knitting machine, the multi-component elements are fabricated ready-made, without requiring any postprocessing.

By combining regular yarn with electrically conductive and resistive yarn, we can create force sensors that are made of 100% textile material. The resulting material is able to detect arbitrary external force, such as strain and pressure, and can be fabricated using a variety of knit structures and patterns.

We have countless ideas of what an ideal future could look like but often fall short in bringing those to the real world. In fiction, the preconditions upon which a utopia can be built are malleable, making any imagined future a possible scenario.What if we would create a desirable future with the same creative audacity of fiction but with the preconditions and rules of the real world?

We’ve invited a local improvisational performer Flo Amar to make an impro theatre about robots. Our brief was “People who act as people interact with people who act as technology.”

Knitting hardware and software are highly optimized for mass-production and require considerable operating experience and training. We develop tools and scripts for simplifying access to knitting technology for novices, DIY tinkerers, and HCI researchers that come from an CS area, like ourselves. Building on-top of pre-existing platforms, we focus on generating knit that can serve as UI devices.

In BAMBI, we use camera-equipped UAVs to capture video data of infrared and visible light to monitor animals in the wild. The objective is to analyze wildlife populations to identify impending ecological issues, such as overpopulation or loss of biodiversity, on a long-term scale. We apply Machine Learning models for automatic identification and analysis. Therefore, we explore interaction methods for handling multi-modal data. In particular, we are investigating methods for annotation and reporting.

Interaction with smart textiles has huge potential for making interacting with new technology more seamless, as textiles are comfortable and also quite familiar. In this work, we explore the design space for communicating surface gestures to users of textile interfaces by experimenting with the interfaces’ physical design and affordances. We reflect on a number of samples that were created based on the aspects of design, fabrication, and sensing.

TexYZ is a method for creating capacitive multitouch sensor on textile basis, which can be implemented using a commodity embroidery machine. We use publicly available enameled copper wire as a bobbin thread to form a matrix of capacitors on the backside of the fabric, while the front threads are holding the patterns in place. The textile’s front has an appealing haptic quality and can act as a visual cue for interaction.TexYZ is using a mutual capacitive sensing technique, meaning it can handle multitouch without showing ghosting effects, which is a central limitation to previous textile touch sensors.

Sonoflex is a flexible, textile dynamic speaker made without using a permanent magnet. Our design consists of two flat spiral coils, stacked on top of each other, and is based on an isolated, thin enameled copper wire. Our approach allows for thin, lightweight, and flexible embroidered devices and does not require high voltage as in electrostatic speakers. We show how Sonoflex can be designed and fabricated, and evaluate its acoustic properties with respect to manufacturing parameters, such as size, turn counts, turn spacing, and substrate materials. We are able to produce audible sound with a broad frequency range of 1.5 kHz – 20 kHz with a 50 mm diameter implementation.

Knitted RESi is a knit textile capable of sensing external stress, via a force sensing technical yarn. It can sense pressure, stretch, and any other deformation. Our demonstrator provides live feedback and visualization of the sensor readings. The textile is highly durable and strikingly responsive with little signal-to-noise ratio. It features inherent favorable properties of textiles, e.g. it is lightweight, highly flexible, comfortable on skin, and breathable.

Humans tend to touch surfaces out of curiosity about their tactile properties and persist in case of a pleasant response. An example is petting a cat. It looks pleasant to touch and when the cat’s response is positive (by purring) and a causal relation to the interaction is obvious, the interaction carries on. This connection is also true for the opposite, as rejecting feedback is uninviting. Inspired by observations like these, the installation The Sound of Textile links textiles’ haptic properties to sound, in order to engage with the visitor and generate an effect similar to synesthesia. The installation demonstrates seven textiles with unique tactile properties and according sonic experiences. The objective is to explore humans’ reactions to different features and their combinations.

We create pressure sensors, which are made of 100% textile components and work just like common FSR (Force-Sensing Resistor) sensors. Using an ordinary embroidery machine, our sensors can be rapidly applied to numerous pre-existing fabrics, in arbitrary shape and scale. Just as print FSR, they are straightforward to use in combination with basic readout electronics. We also developed a custom PCB, which can be attached directly onto the augmented fabric. Even wiring for connecting the PCB to sensors is done using textile materials, by embroidering the respective circuit paths. We can also link several sensors for building pressure sensitive configurations with high flexibility.

As smart textiles are becoming more present in our lives, investigating and designing textile interfaces has started getting more and more attention. Still, very little research has been done on how to design interactive elements for non-wearable textile interfaces for the best recognition, perception, and interaction. In this paper, we present initial assumptions for designing such interfaces, which we derived from working intensively with our partners from the industry. These have been further explored with experts from the field during interviews, and finally tested in a user study. As a conclusion of the study, we define five design recommendations for textile interfaces and present several prototypes that demonstrate them in practice.