Xuanyou Liu
Home
CHI 2024 · Full Paper

TacTex

A textile interface with seamlessly-integrated electrodes for high-resolution electrotactile stimulation

Hongnan Lin, Xuanyou Liu, Shengsheng Jiang, Qi Wang, Ye Tao, Guanyun Wang, Wei Sun, Teng Han, Feng Tian

CHI 2024

Three photos of the TacTex textile. Left: a person touches a large woven panel lying beside a laptop. Top right: a close-up of the weave with conductive and non-conductive yarns interlaced. Bottom right: a person rests both hands and forearms on a fabric sheet with woven electrode regions.

Overview

We developed a fabric that works as a tactile display: it renders patterns you can feel directly on the skin and senses where you touch it. The display is electrotactile: tiny pulses stimulate the touch receptors under the skin, so soft cloth produces rich, controllable sensations.

Dense electrode grids normally fight fabric, so prior interfaces attached patches on afterwards. TacTex weaves the electrodes into the cloth in a single run on a standard loom: the finished textile is already a display and still feels like everyday fabric.

two jobs, one fabricpulses felt on the skintouch position sensedcrossing yarnsare the electrodes
A fabric that is also a display. Conductive yarns are woven into an electrode grid; the same grid delivers pulses and senses where a finger is.
512 × 512stimulation and sensing resolution
2 mmelectrode spacing, as small as
4sensation qualities: pricking, caressing, tapping, pressing
10 + 10static and dynamic patterns

Creating touch with electricity

Soft cloth renders patterns you can feel by stimulating receptors under the skin, with nothing physically moving.

A diagram of a cross section of the skin, titled a tactile pixel on the electrotactile display. Two electrodes are shown labeled with plus and minus correspondingly. An arrow pointing from plus electrode goes to the minus electrode. The arrow is labeled with a lightning icon and the word pulse current. The arrow intersects with a drawing of a tactile receptor, with more receptors drawn nearby.
Each “tactile pixel” on the display consists of a pair of electrodes; in TacTex, these are two conductive yarns in the weave. Whenever the pixel is supposed to be felt, our system passes tiny current pulses between the electrodes, which stimulate the tactile receptors under the skin, causing a sense of slight touch, with nothing physically moving.
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; shifting the active pixels over time draws a dynamic pattern, such as a line sweeping across the palm. Changing the pulses themselves changes the quality of the sensation: pricking, caressing, tapping, or pressing.

1The tactile pixel

Each “tactile pixel” on the display consists of a pair of electrodes; in TacTex, these are two conductive yarns in the weave. Whenever the pixel is supposed to be felt, our system passes tiny current pulses between the electrodes, which stimulate the tactile receptors under the skin, causing a sense of slight touch, with nothing physically moving.

A diagram of a cross section of the skin, titled a tactile pixel on the electrotactile display. Two electrodes are shown labeled with plus and minus correspondingly. An arrow pointing from plus electrode goes to the minus electrode. The arrow is labeled with a lightning icon and the word pulse current. The arrow intersects with a drawing of a tactile receptor, with more receptors drawn nearby.
A tactile pixel, as drawn in the paper.
static patterna thin vertical linedynamic patternthe line sweeps across the fabric
Static and dynamic patterns on the weave.

2Patterns and qualities

Activating several pixels together draws a static pattern, such as a thin vertical line; shifting the active pixels over time draws a dynamic pattern, such as a line sweeping across the palm. Changing the pulses themselves changes the quality of the sensation: pricking, caressing, tapping, or pressing.

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.
Lines running through the weave under a fingertip.

Implementation

Three parts make this work: the woven textile that carries the electrodes, the circuit that drives and reads them, and the connection between fabric and electronics.

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 yarns run lengthwise and crosswise through a multi-layer weave, and non-conductive yarns sit between them so the two directions never short-circuit inside the fabric, while both stay exposed on the side that meets the skin. That shared electrode grid drives both electrotactile stimulation and capacitive touch sensing. Electrode width and spacing are set simply by yarn counts, down to 2 mm. The same structure also supports double-sided displays, insulated backs, and decorative colours.
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 pixel receives pulses at each moment. Each pulse is very short, so the boards can cycle through many pixels fast enough that separate spots feel simultaneous. In the gaps between pulses, the same boards measure the voltage on the yarns to detect where a finger is resting. This is how stimulation and touch 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. Fabric cannot be soldered, so the electrode yarns at the edge of the textile are clamped against a flexible printed connector with a conductive adhesive, giving each yarn its own clean channel to the driving board. Because the electrodes are continuous lines across the cloth rather than individual pads, a 512×512-point display only needs 512 + 512 connections instead of one per point.

1The textile

Conductive yarns run lengthwise and crosswise through a multi-layer weave, and non-conductive yarns sit between them so the two directions never short-circuit inside the fabric, while both stay exposed on the side that meets the skin. That shared electrode grid drives both electrotactile stimulation and capacitive touch sensing. Electrode width and spacing are set simply by yarn counts, down to 2 mm. The same structure also supports double-sided displays, insulated backs, and decorative colours.

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.
The weave structure, with electrode size and spacing set by yarn counts.

2The circuit

Injectionstimulating pulses pass through the skin and are feltSensinga short pulse, read at the sampling resistortextile electrodesH1L1H2L2H3L3every other electrode stays at high impedancefelt as touchimperceptible to the fingeradjustable high-voltageconstant-current supplysampling resistorswitch Hswitch LskinampAD8479ADCADS7044STM32comparereading = referencenot touchedreading differs from referencetouched
Simplified schematic of the driving circuit, animated.
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.
The driving board, part of a custom stack of power, switch, and sensing boards.

One circuit, two jobs. TacTex senses touch with the hardware that stimulates the skin: the supply, the switches, and the electrodes, shared in turns (time-division multiplexing).

While injecting, pulses pass through an activated pair of electrodes and the finger, and are felt. While sensing, a short pulse passes through a sampling resistor, and the sensing board compares its voltage with a reference: a match means the pair is not touched, a difference means it is.

Each electrode switch can tie its electrode to high voltage, to ground, or to high impedance; the schematic draws only the two switches of the active pair.

3The connection

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.
The flexible printed connector, and the heat press that clamps it onto the textile.

Fabric cannot be soldered, so the electrode yarns at the edge of the textile are clamped against a flexible printed connector with a conductive adhesive, giving each yarn its own clean channel to the driving board. Because the electrodes are continuous lines across the cloth rather than individual pads, a 512×512-point display only needs 512 + 512 connections instead of one per point.

User study

We evaluated whether people can actually read the display: tell the sensation qualities apart, and recognise the static and dynamic patterns.

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. Participants felt each stimulus on the textile without seeing it, then identified which quality or pattern it was and judged whether pairs of sensations were the same or different.
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 adjacent lines, which is what sets the practical resolution of the display.

1What 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.

A needle held just above a fingertip, illustrating pricking.
Pricking
Two fingers brushing against each other, illustrating caressing.
Caressing
A fingertip hovering just above a surface, illustrating tapping.
Tapping
A fingertip pressing down onto a surface, illustrating pressing.
Pressing
The four sensation qualities, illustrated.
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.
The study setup, with a close-up of a fingertip on the textile.

2The protocol

Participants felt each stimulus on the textile without seeing it, then identified which quality or pattern it was and judged whether pairs of sensations were the same or different.

Twelve healthy adults took part, none with prior experience of electrotactile stimulation. In each of three tests (qualities, static patterns, dynamic patterns), every stimulus was presented six times in random order. Afterwards they rated the electrode-induced pain at 2.75 on average (SD 1.60) on a scale from 0, no pain, to 10.

3What we found

12adults took part, none with prior electrotactile experience
0.78accuracy telling the four sensation qualities apart
0.88accuracy recognising the ten static patterns
0.75accuracy recognising the ten dynamic patterns
Sensation qualitiesaccuracy 0.7805tappingtappingprickingprickingpressingpressingcaressingcaressingPresented tapping, reported tapping: 61 of 7261Presented tapping, reported pricking: 6 of 726Presented tapping, reported pressing: 0 of 720Presented tapping, reported caressing: 5 of 725Presented pricking, reported tapping: 10 of 7210Presented pricking, reported pricking: 57 of 7257Presented pricking, reported pressing: 0 of 720Presented pricking, reported caressing: 5 of 725Presented pressing, reported tapping: 1 of 711Presented pressing, reported pricking: 1 of 711Presented pressing, reported pressing: 61 of 7161Presented pressing, reported caressing: 8 of 718Presented caressing, reported tapping: 3 of 723Presented caressing, reported pricking: 9 of 729Presented caressing, reported pressing: 15 of 7215Presented caressing, reported caressing: 45 of 7245rows: presented, columns: reportedreported correctly (deeper = more of that row)reported as another onelargest confusion

Rows are the pattern presented, columns the pattern reported. The purple diagonal means responses mostly landed on what was actually rendered; the red cells off the diagonal are the mistakes.

Caressing was the hardest quality: 15 of its 72 presentations were reported as pressing. Pricking and tapping were never reported as pressing. That fits how participants described them: pricking and tapping as discrete sensations, caressing and pressing as continuous.

Static patternsaccuracy 0.8792, numbers as in Figure 6D of the paper556677889910101111121213131414Presented pattern 5, reported pattern 5: 70 of 7270Presented pattern 5, reported pattern 6: 1 of 721Presented pattern 5, reported pattern 7: 0 of 720Presented pattern 5, reported pattern 8: 0 of 720Presented pattern 5, reported pattern 9: 0 of 720Presented pattern 5, reported pattern 10: 0 of 720Presented pattern 5, reported pattern 11: 1 of 721Presented pattern 5, reported pattern 12: 0 of 720Presented pattern 5, reported pattern 13: 0 of 720Presented pattern 5, reported pattern 14: 0 of 720Presented pattern 6, reported pattern 5: 1 of 891Presented pattern 6, reported pattern 6: 83 of 8983Presented pattern 6, reported pattern 7: 0 of 890Presented pattern 6, reported pattern 8: 1 of 891Presented pattern 6, reported pattern 9: 3 of 893Presented pattern 6, reported pattern 10: 0 of 890Presented pattern 6, reported pattern 11: 1 of 891Presented pattern 6, reported pattern 12: 0 of 890Presented pattern 6, reported pattern 13: 0 of 890Presented pattern 6, reported pattern 14: 0 of 890Presented pattern 7, reported pattern 5: 0 of 720Presented pattern 7, reported pattern 6: 1 of 721Presented pattern 7, reported pattern 7: 69 of 7269Presented pattern 7, reported pattern 8: 0 of 720Presented pattern 7, reported pattern 9: 0 of 720Presented pattern 7, reported pattern 10: 0 of 720Presented pattern 7, reported pattern 11: 1 of 721Presented pattern 7, reported pattern 12: 1 of 721Presented pattern 7, reported pattern 13: 0 of 720Presented pattern 7, reported pattern 14: 0 of 720Presented pattern 8, reported pattern 5: 0 of 710Presented pattern 8, reported pattern 6: 0 of 710Presented pattern 8, reported pattern 7: 0 of 710Presented pattern 8, reported pattern 8: 70 of 7170Presented pattern 8, reported pattern 9: 1 of 711Presented pattern 8, reported pattern 10: 0 of 710Presented pattern 8, reported pattern 11: 0 of 710Presented pattern 8, reported pattern 12: 0 of 710Presented pattern 8, reported pattern 13: 0 of 710Presented pattern 8, reported pattern 14: 0 of 710Presented pattern 9, reported pattern 5: 1 of 901Presented pattern 9, reported pattern 6: 1 of 901Presented pattern 9, reported pattern 7: 0 of 900Presented pattern 9, reported pattern 8: 3 of 903Presented pattern 9, reported pattern 9: 84 of 9084Presented pattern 9, reported pattern 10: 1 of 901Presented pattern 9, reported pattern 11: 0 of 900Presented pattern 9, reported pattern 12: 0 of 900Presented pattern 9, reported pattern 13: 0 of 900Presented pattern 9, reported pattern 14: 0 of 900Presented pattern 10, reported pattern 5: 0 of 720Presented pattern 10, reported pattern 6: 0 of 720Presented pattern 10, reported pattern 7: 0 of 720Presented pattern 10, reported pattern 8: 0 of 720Presented pattern 10, reported pattern 9: 1 of 721Presented pattern 10, reported pattern 10: 71 of 7271Presented pattern 10, reported pattern 11: 0 of 720Presented pattern 10, reported pattern 12: 0 of 720Presented pattern 10, reported pattern 13: 0 of 720Presented pattern 10, reported pattern 14: 0 of 720Presented pattern 11, reported pattern 5: 0 of 720Presented pattern 11, reported pattern 6: 3 of 723Presented pattern 11, reported pattern 7: 0 of 720Presented pattern 11, reported pattern 8: 0 of 720Presented pattern 11, reported pattern 9: 1 of 721Presented pattern 11, reported pattern 10: 0 of 720Presented pattern 11, reported pattern 11: 56 of 7256Presented pattern 11, reported pattern 12: 3 of 723Presented pattern 11, reported pattern 13: 7 of 727Presented pattern 11, reported pattern 14: 2 of 722Presented pattern 12, reported pattern 5: 1 of 711Presented pattern 12, reported pattern 6: 0 of 710Presented pattern 12, reported pattern 7: 0 of 710Presented pattern 12, reported pattern 8: 0 of 710Presented pattern 12, reported pattern 9: 0 of 710Presented pattern 12, reported pattern 10: 1 of 711Presented pattern 12, reported pattern 11: 15 of 7115Presented pattern 12, reported pattern 12: 49 of 7149Presented pattern 12, reported pattern 13: 1 of 711Presented pattern 12, reported pattern 14: 4 of 714Presented pattern 13, reported pattern 5: 1 of 721Presented pattern 13, reported pattern 6: 1 of 721Presented pattern 13, reported pattern 7: 1 of 721Presented pattern 13, reported pattern 8: 2 of 722Presented pattern 13, reported pattern 9: 10 of 7210Presented pattern 13, reported pattern 10: 0 of 720Presented pattern 13, reported pattern 11: 2 of 722Presented pattern 13, reported pattern 12: 0 of 720Presented pattern 13, reported pattern 13: 53 of 7253Presented pattern 13, reported pattern 14: 2 of 722Presented pattern 14, reported pattern 5: 0 of 720Presented pattern 14, reported pattern 6: 0 of 720Presented pattern 14, reported pattern 7: 0 of 720Presented pattern 14, reported pattern 8: 0 of 720Presented pattern 14, reported pattern 9: 0 of 720Presented pattern 14, reported pattern 10: 0 of 720Presented pattern 14, reported pattern 11: 0 of 720Presented pattern 14, reported pattern 12: 4 of 724Presented pattern 14, reported pattern 13: 11 of 7211Presented pattern 14, reported pattern 14: 57 of 7257rows: presented, columns: reportedDynamic patternsaccuracy 0.7492, numbers as in Figure 6E of the paper1515161617171818191920202121222223232424Presented pattern 15, reported pattern 15: 60 of 7160Presented pattern 15, reported pattern 16: 2 of 712Presented pattern 15, reported pattern 17: 3 of 713Presented pattern 15, reported pattern 18: 0 of 710Presented pattern 15, reported pattern 19: 1 of 711Presented pattern 15, reported pattern 20: 0 of 710Presented pattern 15, reported pattern 21: 0 of 710Presented pattern 15, reported pattern 22: 0 of 710Presented pattern 15, reported pattern 23: 3 of 713Presented pattern 15, reported pattern 24: 2 of 712Presented pattern 16, reported pattern 15: 0 of 690Presented pattern 16, reported pattern 16: 57 of 6957Presented pattern 16, reported pattern 17: 0 of 690Presented pattern 16, reported pattern 18: 3 of 693Presented pattern 16, reported pattern 19: 2 of 692Presented pattern 16, reported pattern 20: 0 of 690Presented pattern 16, reported pattern 21: 2 of 692Presented pattern 16, reported pattern 22: 1 of 691Presented pattern 16, reported pattern 23: 0 of 690Presented pattern 16, reported pattern 24: 4 of 694Presented pattern 17, reported pattern 15: 0 of 690Presented pattern 17, reported pattern 16: 5 of 695Presented pattern 17, reported pattern 17: 52 of 6952Presented pattern 17, reported pattern 18: 2 of 692Presented pattern 17, reported pattern 19: 0 of 690Presented pattern 17, reported pattern 20: 6 of 696Presented pattern 17, reported pattern 21: 0 of 690Presented pattern 17, reported pattern 22: 0 of 690Presented pattern 17, reported pattern 23: 3 of 693Presented pattern 17, reported pattern 24: 1 of 691Presented pattern 18, reported pattern 15: 0 of 670Presented pattern 18, reported pattern 16: 0 of 670Presented pattern 18, reported pattern 17: 2 of 672Presented pattern 18, reported pattern 18: 55 of 6755Presented pattern 18, reported pattern 19: 0 of 670Presented pattern 18, reported pattern 20: 0 of 670Presented pattern 18, reported pattern 21: 0 of 670Presented pattern 18, reported pattern 22: 1 of 671Presented pattern 18, reported pattern 23: 3 of 673Presented pattern 18, reported pattern 24: 6 of 676Presented pattern 19, reported pattern 15: 1 of 691Presented pattern 19, reported pattern 16: 1 of 691Presented pattern 19, reported pattern 17: 0 of 690Presented pattern 19, reported pattern 18: 1 of 691Presented pattern 19, reported pattern 19: 58 of 6958Presented pattern 19, reported pattern 20: 2 of 692Presented pattern 19, reported pattern 21: 2 of 692Presented pattern 19, reported pattern 22: 1 of 691Presented pattern 19, reported pattern 23: 3 of 693Presented pattern 19, reported pattern 24: 0 of 690Presented pattern 20, reported pattern 15: 2 of 692Presented pattern 20, reported pattern 16: 0 of 690Presented pattern 20, reported pattern 17: 1 of 691Presented pattern 20, reported pattern 18: 4 of 694Presented pattern 20, reported pattern 19: 2 of 692Presented pattern 20, reported pattern 20: 56 of 6956Presented pattern 20, reported pattern 21: 0 of 690Presented pattern 20, reported pattern 22: 3 of 693Presented pattern 20, reported pattern 23: 1 of 691Presented pattern 20, reported pattern 24: 0 of 690Presented pattern 21, reported pattern 15: 0 of 690Presented pattern 21, reported pattern 16: 3 of 693Presented pattern 21, reported pattern 17: 0 of 690Presented pattern 21, reported pattern 18: 0 of 690Presented pattern 21, reported pattern 19: 4 of 694Presented pattern 21, reported pattern 20: 1 of 691Presented pattern 21, reported pattern 21: 53 of 6953Presented pattern 21, reported pattern 22: 4 of 694Presented pattern 21, reported pattern 23: 0 of 690Presented pattern 21, reported pattern 24: 4 of 694Presented pattern 22, reported pattern 15: 1 of 681Presented pattern 22, reported pattern 16: 0 of 680Presented pattern 22, reported pattern 17: 4 of 684Presented pattern 22, reported pattern 18: 1 of 681Presented pattern 22, reported pattern 19: 3 of 683Presented pattern 22, reported pattern 20: 1 of 681Presented pattern 22, reported pattern 21: 1 of 681Presented pattern 22, reported pattern 22: 54 of 6854Presented pattern 22, reported pattern 23: 3 of 683Presented pattern 22, reported pattern 24: 0 of 680Presented pattern 23, reported pattern 15: 4 of 654Presented pattern 23, reported pattern 16: 0 of 650Presented pattern 23, reported pattern 17: 0 of 650Presented pattern 23, reported pattern 18: 8 of 658Presented pattern 23, reported pattern 19: 0 of 650Presented pattern 23, reported pattern 20: 1 of 651Presented pattern 23, reported pattern 21: 4 of 654Presented pattern 23, reported pattern 22: 3 of 653Presented pattern 23, reported pattern 23: 38 of 6538Presented pattern 23, reported pattern 24: 7 of 657Presented pattern 24, reported pattern 15: 0 of 710Presented pattern 24, reported pattern 16: 2 of 712Presented pattern 24, reported pattern 17: 6 of 716Presented pattern 24, reported pattern 18: 1 of 711Presented pattern 24, reported pattern 19: 1 of 711Presented pattern 24, reported pattern 20: 3 of 713Presented pattern 24, reported pattern 21: 0 of 710Presented pattern 24, reported pattern 22: 1 of 711Presented pattern 24, reported pattern 23: 13 of 7113Presented pattern 24, reported pattern 24: 44 of 7144rows: presented, columns: reportedreported correctly (deeper = more of that row)reported as another onelargest confusion
Mistakes happen 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) were hardest, with the lowest balanced accuracies (0.7690 and 0.7903); pattern 24 was reported as 23 in 13 of 71. These off-diagonal cells are what set the practical resolution of the display. 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}
}