TacTex: A Textile Interface with Seamlessly-Integrated Electrodes for High-Resolution Electrotactile Stimulation

Electrotactile stimulation offers a promising path for high-resolution wearable haptics but has traditionally struggled with integration into soft, everyday textiles. We present TacTex, a textile interface that seamlessly integrates high-density haptic feedback and touch sensing. By employing a novel multi-layer woven structure that separates conductive electrodes with non-conductive yarns, TacTex achieves a sensing and actuation resolution of 512 × 512 with electrode spacing as small as 2 mm. The system includes a custom driving board that enables precise spatial and temporal control of electrical stimuli while simultaneously monitoring voltage changes for touch tracking. Our technical evaluation and user studies demonstrate that TacTex can render a wide range of haptic effects, including complex static and dynamic patterns, effectively bringing high-fidelity haptic feedback to everyday clothing.

The textile. A dense woven electrode array giving both stimulation and touch tracking at 512×512 resolution, from linear electrodes spaced as closely as 2 mm.
The driving system. Spatial and temporal control of the stimuli while simultaneously monitoring voltage, so touch tracking never disturbs the haptics.
The vocabulary. Four sensation qualities, ten static and ten dynamic patterns, all distinguishable in user evaluations.

Electrotactile stimulation

Electrotactile stimulation injects current through surface electrodes to stimulate afferent nerve endings in the skin, which produces sensations such as touch, vibration, pressure and spatial patterns. TacTex uses linear electrodes, which simplifies the electrodes, the wiring and the driving circuit.

A close-up of the woven textile with a fingertip touching it. Overlaid red and blue lines run through the weave, tracing the conductive yarns that form patterns on the fabric. — Activating several pixels together draws a **static pattern**, such as a thin vertical line; updating the active pixels over time draws a **dynamic pattern**, such as a line sweeping across the fabric. The **quality** of the sensation (pricking, caressing, tapping or pressing) is set by the pixel properties and the burst parameters.

Tactile pixel

Each tactile pixel is a pair of parallel electrodes; in TacTex these are two conductive yarns in the weave. Current pulses between the two yarns stimulate afferent nerve endings in the skin and are felt as a localised touch. Pixels activated within the same burst period are perceived as activated together.

Static and dynamic patterns on the weave.

Patterns and qualities

Activating several pixels together draws a static pattern, such as a thin vertical line; updating the active pixels over time draws a dynamic pattern, such as a line sweeping across the fabric. The quality of the sensation (pricking, caressing, tapping or pressing) is set by the pixel properties and the burst parameters.

Implementation

TacTex consists of a woven electrode array, a driving circuit that stimulates the electrodes and reads touch through the same hardware, and a connection between the two.

The weave structure. Top left: a 3D diagram of the multi-layer weave unit with labelled warp and weft yarns. Below: variants showing how yarn counts set electrode size and spacing. Right: a photograph of the woven fabric with a 1 cm scale bar and a rendering of the weave overlaid. — **Textile design.** Conductive and non-conductive yarns are woven in one run, on a rapier loom in our evaluation, into a two-layer structure. Warp electrodes are separated from each other by non-conductive warp yarns and weft electrodes by non-conductive weft yarns, and the two sets are kept apart by non-conductive weft yarns on the bottom layer. Electrode size follows from the number of conductive yarns per electrode and electrode spacing from the number of non-conductive yarns between electrodes, down to **2 mm**. Closer spacing raises electrode density but also the risk of short circuits; the paper reports short-circuit rates for several weave variants and mitigates the risk with additional layers. The same electrodes are used for stimulation and for touch tracking.

The driving board: a green circuit board with a microcontroller, a 200 volt input and two 200 volt outputs, data and power connectors, and rows of sockets that route stimulation and sensing to the textile's electrodes. — **Driving circuit.** A custom stack of power, switch and sensing boards selects which electrode pair receives pulses at each moment. Pulses are short, so the pairs can be cycled fast enough that pixels activated within the same burst period feel simultaneous. Between stimulation pulses, the same hardware **measures a voltage to detect touch**, so stimulation and sensing share one surface.

The fabric-to-board connection. Top left: a diagram of edge electrodes bonding to pitch-matched pads. Bottom left: a close-up of the flexible printed connector with fanned-out golden traces. Right: a heat press clamping the connector onto the woven textile. — **Connection.** The electrodes at the textile edge are joined to contact pads on a custom **flexible printed cable** (polyimide and copper) whose pad pitch matches the textile, using an anisotropic conductive adhesive (ACA). The adhesive conducts only in the vertical direction between stacked contacts, so each electrode gets its own channel to the driving board without adding rigid or bulky elements. Because the electrodes are continuous lines across the cloth, a 512×512 array needs 512 + 512 connections instead of one per point.

Textile design

Conductive and non-conductive yarns are woven in one run, on a rapier loom in our evaluation, into a two-layer structure. Warp electrodes are separated from each other by non-conductive warp yarns and weft electrodes by non-conductive weft yarns, and the two sets are kept apart by non-conductive weft yarns on the bottom layer. Electrode size follows from the number of conductive yarns per electrode and electrode spacing from the number of non-conductive yarns between electrodes, down to 2 mm. Closer spacing raises electrode density but also the risk of short circuits; the paper reports short-circuit rates for several weave variants and mitigates the risk with additional layers. The same electrodes are used for stimulation and for touch tracking.

Driving circuit

Simplified schematic of the driving circuit, animated.

Stimulation and touch tracking use the same hardware: the power supply, the switches and the electrodes, shared in turns by time-division multiplexing. The circuit consists of an adjustable high-voltage constant-current supply, a sensing board, and 16 sets of controllers paired with tri-state switches. Each switch can connect its electrode to a high voltage, to ground, or leave it at high impedance; the schematic shows only the two switches of the active pair.

During stimulation, pulses pass through an activated pair of electrodes and the finger. During sensing, a short pulse passes through a sampling resistor, and the sensing board compares the measured voltage with a reference value. If the two are equal, the pair is not touched; otherwise it is.

Connection to the driving board

The electrodes at the textile edge are joined to contact pads on a custom flexible printed cable (polyimide and copper) whose pad pitch matches the textile, using an anisotropic conductive adhesive (ACA). The adhesive conducts only in the vertical direction between stacked contacts, so each electrode gets its own channel to the driving board without adding rigid or bulky elements. Because the electrodes are continuous lines across the cloth, a 512×512 array needs 512 + 512 connections instead of one per point.

User study

Experiment 1 measured how accurately participants could identify the four sensation qualities, the ten static patterns and the ten dynamic patterns rendered by TacTex.

The stimulus set. Top: photographs of four sensation qualities delivered to a fingertip. Below: grids showing static patterns, such as single vertical or horizontal lines, and dynamic patterns rendered as sequences of bars sweeping across the array. — **What we render.** The display produces **four sensation qualities** (pricking, caressing, tapping, and pressing) plus **ten static patterns**, such as a thin vertical line, and **ten dynamic patterns**, such as a line sweeping across the fabric.

The study setup: a participant rests their hand on the electrode board, connected by a ribbon cable to the drive stack and a laptop showing a practice trial; an inset shows a fingertip touching the woven textile. — **Study protocol.** Twelve participants **identified** the sensation quality or pattern of each stimulus, presented six times in random order.

Two stimulus-perception matrices, one for static patterns and one for dynamic patterns. Each is a ten by ten grid where rows are the pattern presented and columns the pattern reported. Bright cells run along the diagonal, showing that the pattern participants reported was usually the one presented. — **What we found.** Rows are the pattern presented, columns the pattern reported. The **bright diagonal** means responses mostly landed on the pattern actually rendered; the off-diagonal cells cluster among similar patterns.

Stimuli

The stimulus set has four sensation qualities, ten static patterns and ten dynamic patterns. Pricking is a discrete, sharp and shallow sensation; caressing is continuous, blurry and shallow; tapping is discrete, blurry and deep; pressing is continuous, blurry and deep. Dynamic patterns update the active pixels frame by frame and repeat every 2 seconds.

A needle held just above a fingertip, illustrating pricking. — The four sensation qualities, illustrated.

Two fingers brushing against each other, illustrating caressing. — The four sensation qualities, illustrated.

Procedure

Twelve healthy adults (6 female, mean age 24.5 years, SD 3.68) took part after giving informed consent; none had prior experience with electrotactile stimulation. Each participant's sensation threshold and maximum comfortable level were calibrated first, and the stimulus current was set to the threshold plus 80% of the range up to the maximum. After practice trials, participants completed three tests: sensation qualities, static patterns and dynamic patterns. Within a test, each item was presented six times in random order, and participants chose the sensation or pattern they perceived from the options shown on a computer, with no time limit. Afterwards they rated electrode-induced pain on a 0 to 10 numeric rating scale (mean 2.75, SD 1.60).

Results

Test	Items	Accuracy	Per-class balanced accuracy
Sensation qualities	4	0.78	0.77 to 0.89
Static patterns	10	0.88	0.84 to 0.99
Dynamic patterns	10	0.75	0.77 to 0.92

Identification accuracy averaged over the 12 participants; every item was presented six times (paper Section 4.2.2).

Rows give the quality presented and columns the quality reported. The diagonal (purple) holds correct responses and the off-diagonal cells (red) are confusions.

Caressing had the lowest balanced accuracy of the four qualities (0.77): 15 of its 72 presentations were reported as pressing. Pricking and tapping were never reported as pressing. In the interviews, participants described pricking and tapping as discrete sensations and caressing and pressing as continuous.

Stimulus-perception matrices for the static and dynamic patterns. Confusions occur mostly between similar patterns: five vertical lines (pattern 12) were reported as three vertical lines (pattern 11) in 15 of 71 presentations. Among the dynamic patterns, the two grid animations (23 and 24) had the lowest balanced accuracies (0.7690 and 0.7903), and pattern 24 was reported as 23 in 13 of 71. Hover a cell for its count.

Team

This is a collaboration between the Institute of Software (Chinese Academy of Sciences), Zhejiang University, and Hangzhou City University by:

Together with Shengsheng Jiang, Qi Wang, Wei Sun, and Feng Tian.

Publication

This work is published at ACM CHI 2024 (see paper via DOI). Authors’ print is available here (PDF).

Lin, H., Liu, X., Jiang, S., Wang, Q., Tao, Y., Wang, G., Sun, W., Han, T., & Tian, F. (2024). TacTex: A Textile Interface with Seamlessly-Integrated Electrodes for High-Resolution Electrotactile Stimulation. In Proceedings of the CHI Conference on Human Factors in Computing Systems (CHI ’24). https://doi.org/10.1145/3613904.3642873

@inproceedings{lin2024tactex,
  title={TacTex: A Textile Interface with Seamlessly-Integrated Electrodes for High-Resolution Electrotactile Stimulation},
  author={Lin, Hongnan and Liu, Xuanyou and Jiang, Shengsheng and Wang, Qi and Tao, Ye and Wang, Guanyun and Sun, Wei and Han, Teng and Tian, Feng},
  booktitle={Proceedings of the 2024 CHI Conference on Human Factors in Computing Systems},
  year={2024},
  doi={10.1145/3613904.3642873}
}

TacTex

Electrotactile stimulation

Implementation

User study

Team

Publication

TacTex

Tiny pulses stimulate the skin, so nothing has to move.

A pair of electrodes

Many pixels, one shape

Change the pulses, change the feel

Weave the electrodes into the cloth, in one run on a standard loom.

A multi-layer weave keeps the two electrode directions apart.

No short circuit

Set width and spacing

More than a patch

One circuit, two jobs: inject pulses, then sense touch.

Pulses are felt

Voltage tells where

In turns

A custom board stack drives the cloth through a flexible connector.

Twelve adults felt each stimulus, then chose what they had perceived.

Twelve healthy adults

Qualities, static, dynamic

Pain rated 2.75 on average

The four qualities were told apart with 0.78 accuracy.

Static patterns reached 0.88, dynamic patterns 0.75.

The cloth is the display, and the sensor.

Balanced accuracy of every class, as reported in the paper.

What the participants did, step by step.

Familiarise

Three tests

Pain rating

Abstract

Electrotactile stimulation

Implementation

User study

Team

Publication