Exploring Human-AI Collaboration in E-Textile Design: A Case Study on Flex Sensor Placement for Shoulder Motion Detection DOI: XXXXXXX.XXXXXXX Conference: Make sure to enter the correct conference title from your rights confirmation email; June 03–05, 2018; Woodstock, NY ISBN: 978-1-4503-XXXX-X/2018/06 CCS: Human-centered computing Empirical studies in HCI CCS: Human-centered computing User studies CCS: Human-centered computing Collaborative interaction CCS: Human-centered computing Empirical studies in ubiquitous and mobile computing CCS: Computing methodologies Natural language processing CCS: Computing methodologies Knowledge representation and reasoning CCS: Computer systems organization Sensors and actuators
Zhuchenyang Liu, Yao Zhang, Yalan He, Hilla Paasio, Changyi Li, Guna Semjonova, Yu Xiao · 2026
Read Exploring Human-AI Collaboration in E-Textile Design: A Case Study on Flex Sensor Placement for Shoulder Motion Detection DOI: XXXXXXX.XXXXXXX Conference: Make sure to enter the correct conference title from your rights confirmation email; June 03–05, 2018; Woodstock, NY ISBN: 978-1-4503-XXXX-X/2018/06 CCS: Human-centered computing Empirical studies in HCI CCS: Human-centered computing User studies CCS: Human-centered computing Collaborative interaction CCS: Human-centered computing Empirical studies in ubiquitous and mobile computing CCS: Computing methodologies Natural language processing CCS: Computing methodologies Knowledge representation and reasoning CCS: Computer systems organization Sensors and actuators by Zhuchenyang Liu, Yao Zhang, Yalan He, Hilla Paasio, Changyi Li, Guna Semjonova, Yu Xiao (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2603-13575.txt.
Introduction
Zhuchenyang Liu and co-authors explore human-AI collaboration in e-textile design through a case study. Flex sensors detect joint motion, but placing them well needs anatomy, biomechanics, and textile expertise at once. Their case study targets shoulder motion detection with flex sensors on garments. Three human designers worked in three scenarios: LLMs alone, humans alone, and human-AI collaboration. The least experienced designer improved steadily with AI help and matched the best human-only result. The most experienced designer did worse with collaboration, a surprising split.
Catalog record
| CREATOR | Zhuchenyang Liu, Yao Zhang, Yalan He, Hilla Paasio, Changyi Li, Guna Semjonova, Yu Xiao |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 68568 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2603.13575
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
Exploring Human-AI Collaboration in E-Textile Design: A Case Study on Flex Sensor Placement for Shoulder Motion Detection DOI: XXXXXXX.XXXXXXX Conference: Make sure to enter the correct conference title from your rights confirmation email; June 03–05, 2018; Woodstock, NY ISBN: 978-1-4503-XXXX-X/2018/06 CCS: Human-centered computing Empirical studies in HCI CCS: Human-centered computing User studies CCS: Human-centered computing Collaborative interaction CCS: Human-centered computing Empirical studies in ubiquitous and mobile computing CCS: Computing methodologies Natural language processing CCS: Computing methodologies Knowledge representation and reasoning CCS: Computer systems organization Sensors and actuators Zhuchenyang Liu, Yao Zhang, Yalan He, Hilla Paasio, Changyi Li, Guna Semjonova, Yu Xiao Abstract . Flex sensors are widely used in e-textiles for detecting joint motions and, subsequently, full-body movements. A critical initial step in utilizing these sensors is determining the optimal placement on the body to accurately capture human motions. This task requires a combination of expertise in fields such as anatomy, biomechanics, and textile design, which is seldom found in a single practitioner. Generative AI, such as Large Language Models (LLMs), has recently shown promise in facilitating design. However, to our knowledge, the extent to which LLMs can aid in the e-textile design process remains largely unexplored in the literature. To address this open question, we conducted a case study focusing on shoulder motion detection using flex sensors. We enlisted three human designers to participate in an experiment involving human-AI collaborative design. We examined design efficiency across three scenarios: designs produced by LLMs alone, by humans alone, and through collaboration between LLMs and human designers. Our quantitative and qualitative analyses revealed an intriguing relationship between expertise and outcomes: the least experienced human designer achieved continuous improvement through collaboration, ultimately matching the best performance achieved by humans alone, whereas the most experienced human designer experienced a decline in performance. Additionally, the effectiveness of human-AI collaboration is affected by the granularity of feedback - incremental adjustments outperformed sweeping redesigns - and the level of abstraction, with observation-oriented feedback producing better outcomes than prescriptive anatomical directives. These findings offer valuable insights into the opportunities and challenges associated with human-AI collaborative e-textile design. 1. Introduction A flowchart showing five phases of e-textile sensor layout design with iterative feedback arrows pointing back to Phases 1, 2, and 3. Below, blue (LLM) and orange (Simulation) annotations indicate where each computational approach can assist. Figure 1. Iterative design process of e-textile-based motion capture systems. Each phase is annotated with the domain knowledge it requires. Below the workflow, we map opportunities for AI assistance: LLM-based approaches (blue) can support Phases 1, 4, and 5, while simulation-driven approaches (orange) primarily address Phases 2 and 3. This work focuses on the LLM-assisted stages. A flowchart showing five phases of e-textile sensor layout design with iterative feedback arrows pointing back to Phases 1, 2, and 3. Below, blue (LLM) and orange (Simulation) annotations indicate where each computational approach can assist. Electronic textiles (e-textiles) integrate sensing, computation, and communication directly into fabric, enabling unobtrusive monitoring of human physiology in everyday settings ( 30 ; 18 ; 12 ) . Among the sensing modalities employed in e-textiles, resistive flex sensors have become a popular choice for capturing joint motions and, by extension, full-body movements: bending a flex sensor along a limb or joint produces a resistance change proportional to the deflection angle, which can be mapped to joint kinematics through machine learning models ( 28 ; 20 ; 7 ) . Systems built on this principle have been used for hand gesture recognition ( 16 ) , gait analysis ( 24 ) , and upper-limb rehabilitation monitoring ( 5 ) , demonstrating the breadth of applications that flex-sensor-based motion capture can address. As shown in Fig 1 , the design process for e-textile-based motion capture systems typically involves several iterative steps: designing the sensors and their layout on the human body, fabricating hardware prototypes, collecting sensor readings and corresponding motion measurements, training a prediction model, evaluating performance, and refining the design based on evaluation results ( 19 ; 6 ) . For each specific application, selecting sensor types is often straightforward. For instance, flex sensors are commonly utilized for detecting joint movements, while pressure sensors measure the force applied. Projects not focused on developing new sensor designs generally opt for commercially available or pre-evaluated sensors with clearly defined specifications such as sensitivity, sensing range, and resolution. Given a selection of sensors, determining the optimal placement of sensors on the human body — known as sensor layout design ( 29 ; 16 ) — is a complex optimization challenge. The goal is to maximize the information captured about target movements with a limited number of sensors. Effective sensor placement necessitates integrating knowledge from several fields, including anatomy (identifying which muscles and tendons are involved in the target motion), biomechanics (understanding skin deformation during movement), textile engineering (considering how fabric affects sensor-body coupling), and signal processing (mapping sensor signals to joint angles) ( 9 ; 10 ) . It is challenging for any designer to possess comprehensive expertise in all these areas. Typically, designers adopt a trial-and-error approach, iteratively improving the layout based on evaluation outcomes. This knowledge- and labor-intensive process has spurred efforts to introduce computational assistance to streamline and enhance the design process. Simulation-driven approaches have made notable progress by enabling virtual prototyping through motion and/or sensor simulation, which reduces the cost of physical fabrication and evaluation ( 25 ; 16 ; 26 ; 15 ) . However, in these approaches the design proposals still originate from human designers; the simulation serves primarily to evaluate candidate layouts rather than to generate or refine them. Large Language Models (LLMs) have recently demonstrated potential as cross-domain knowledge integrators in mechanical and physical design tasks ( 22 ; 23 ) . Their ability to synthesize information from various disciplines suggests they could be well-suited to the e-textile design workflow, particularly in sensor layout design. Nevertheless, how and to what extent LLMs can contribute to sensor layout design remains largely unexplored. In this study, we aim to explore the opportunities and challenges associated with leveraging LLMs to assist in the e-textile design process, specifically focusing on sensor layout optimization. We selected shoulder motion detection as our case study due to the shoulder’s complexity as a joint with three degrees of freedom. This choice reflects the shoulder’s role as a representative and clinically significant case in previous e-textile design studies ( 25 ; 16 ; 13 ) . We engaged three human designers with varying levels of expertise in e-textile design — senior, intermediate, and junior — to participate in a user study aimed at addressing the following research questions: RQ1: Can LLMs generate sensor layouts that match or surpass the quality of initial designs produced by human designers across different experience levels? RQ2: When designs generated by LLMs are iteratively refined through human-AI collaboration, how does their quality compare to that of designs produced through a purely human-driven iterative process? RQ3: How do the human designers’ levels of experience, feedback strategies, and interaction patterns affect the effectiveness of human-AI collaboration in the iterative sensor layout optimisation process? We assess design quality based on the motion detection accuracy attained by each design. The scientific contributions of this paper