Toward Motion Capture Textiles: A Batteryless IMU Array Powered and Networked Through a Conductive-Fabric Bus
Listen
Radio episode about this paper
Transcript
Introduction to the show: ident: Electrical Engineering. The day in electrical engineering: signals, systems and the hardware that carries them.
Rosa: We're looking at the paper: "Toward Motion Capture Textiles".
Dev: Here's the gist of the paper. Motion capture is increasingly important in healthcare, 3D Reconstruction rehabilitation, and sports applications This paper proposes a location-independent platform for motion capture, based on an array of IMU (inertial measurement unit) sensor nodes without individual wiring or onboard batteries, enabled by two-dimensional communication and power delivery over conductive fabric System Architecture The proposed system utilizes conductive fabric as a shared two-dimensional medium for both power delivery and data communication The hardware platform involves IMU sensor nodes placed directly on the fabric, with a master node supplying power while collecting posture data The fabric is woven with copper–nickel (Cu–Ni)-plated threads, serving as a two-dimensional shared medium for power and data Communication Protocol A token-passing protocol combined with per-node pipelining enables collision-free and stable data acquisition at 50 Hz Each node fuses the three-axis acceleration and three-axis angular velocity of its IMU into a quaternion using a Madgwick complementary filter on the MCU The protocol assigns transmission rights sequentially, with each node broadcasting a token naming its successor Sensing, computation and communication are pipelined within each node, meaning every transmission is broadcast two slots before its turn Reconstruction Methodology The shape of the fabric is reconstructed by combining IMU orientations with the known grid arrangement of the nodes Adjacent node positions are estimated from fixed-length displacement vectors whose directions follow an intermediate orientation computed by spherical linear interpolation (Slerp) The position of node j is estimated from that of node i as p j = p i + q i, j Lv i, j q-1 i, j A continuous surface follows by cubic Hermite interpolation of the coordinates and the tangents derived from the node orientations Performance Metrics Against a cylinder of known radius, the reconstructed surface deviates by 2.60 mm RMS over the 225 mm span of the array The reconstruction is formulated for the quasi-static regime, and this letter empirically validates it in that regime The system sustains a 50 Hz frame rate across 16 nodes The mean period was 20.206 ms with a standard deviation of 0.127 ms, so the schedule is deterministic to 0.6% Limitations and Future Work Four limitations should be noted, including that each node derives orientation from a filter that uses gravity as attitude reference Because the nodes have no absolute heading reference, yaw drift accumulates, and the flat-state error grew from 1.4 mm to 29.5 mm over 1 min The surface is assumed inextensible, and accuracy should degrade on nondevelopable surfaces and under in-plane strain Long-term use would therefore require periodic recalibration, or a magnetometer to supply the missing reference The paper concludes that motion capture in the full sense will require a reconstruction formulated for dynamic deformation and its characterization against optical motion capture This paper presented a batteryless and wire-free shape reconstruction system using conductive fabric Using the fabric as a shared power and data bus, and superimposing data on DC power via bias-tee circuits, eliminated the need for individual batteries and wiring at each sensor node A token-passing protocol combined with pipelined data processing achieved collision-free real-time data acquisition at 50 Hz in a 4×4 sensor network Against a cylinder of known radius the reconstructed surface deviated by 2.60 mm RMS over the 225 mm span of the array, obtained without a predefined skeletal model The 50 Hz figure is a property of the acquisition system rather than a validated dynamic accuracy The paper also suggests that heterogeneous nodes carrying force or pressure sensors can be added toward zero-moment-point (ZMP) and ground-reaction-force estimation ACKNOWLEDGMENT This research was supported in part by JSPS KAKENHI JP25K03109 and JST CRONOS, Japan, Grant Number JPMJCS25N4 REFERENCES
1: L.
Rosa: Carcreff et al., “Validity, reliability and interpretability of an IMU-based system to measure 3D lower limb kinematics of patients with heterogeneous gait disorders,” Sensors, vol.
Dev: 26, 2026, art.
Rosa: no.
Paper summary: Rosa: So we're looking at this paper called "Toward Motion Capture Textiles: A Batteryless IMU Array Powered and Networked Through a Conductive-Fabric Bus". It basically proposes this location independent platform for motion capture that doesn't need individual wiring or batteries.
Dev: Right, so the core idea is using an array of inertial measurement unit sensor nodes on conductive fabric, powered and communicating through this shared medium.
Rosa: It tackles the problem that conventional vision based systems need cameras and line of sight which limits them in everyday settings. This system aims to be useful outside the lab or in real environments like healthcare or sports applications.
Dev: The paper claims this system uses conductive fabric as a shared two dimensional medium for both power delivery and data communication, eliminating the need for individual batteries at each sensor node.
Taro: But if you're relying on a shared bus for everything, what about reliability when the world gets messy? How does it handle the physical stresses of being woven into cloth?
Rosa: Well, they use copper nickel plated threads for that fabric, and they describe how the master node supplies power to the bottom layer while nodes communicate through conductive pins.
Dev: They also talk about a specific communication protocol—a token passing protocol combined with per node pipelining—which is designed to keep data acquisition collision free and stable at a fifty hertz frame rate <ref:2610.19969#pg2>.
Taro: A token passing system sounds like it could get jammed if the fabric gets too stretched or if nodes drift out of alignment. What happens when that happens?
Rosa: The reconstruction method they use combines the IMU orientations with the known grid arrangement of the nodes, estimating adjacent node positions using fixed length displacement vectors whose directions follow an intermediate orientation computed by spherical linear interpolation, or Slerp.
Dev: And to get a continuous surface, they use cubic Hermite interpolation of the coordinates and tangents derived from those node orientations. That’s how they build the shape when they don't have a predefined skeletal model.
Taro: So what are the actual performance metrics? How good is this fifty hertz reconstruction when you test it against something real, like a cylinder <ref:2610.19969#pg2>?
Rosa: Against a cylinder of known radius, they found that the reconstructed surface deviates by two point six zero millimeters root mean square over the two hundred twenty-five millimeter span of the array.
Dev: That deviation is measured in terms of root mean square error, and they state this reconstruction is formulated for the quasi static regime, which they empirically validated in that specific regime.
Taro: But what about when things get dynamic? The paper mentions limitations like how each node derives its orientation from a filter that uses gravity as an attitude reference.
Rosa: That’s a key point, because they say because the nodes have no absolute heading reference, yaw drift accumulates, and the flat state error grew from one point four millimeters to twenty-nine point five millimeters over one minute.
Dev: And they also flag that the surface is assumed to be inextensible, which means accuracy should degrade on surfaces that aren't developable or under in-plane strain.
Taro: That sounds like a big caveat for real world use, doesn't it? What do you think this means for someone who only listens to the show about motion capture applications?
Rosa: It suggests that while this is great for batteryless and wire free setups, capturing full motion capture in the sense of dynamic deformation will require a reconstruction formulated for dynamic deformation and characterizing that against optical motion capture.
Dev: So, in conclusion, this paper "Toward Motion Capture Textiles: A Batteryless IMU Array Powered and Networked Through a Conductive-Fabric Bus" presents a system that eliminates individual wiring and batteries by using conductive fabric as the shared power and data bus.
Rosa: They achieved collision free real time data acquisition at fifty hertz across a four times four sensor network, and they found that against their test cylinder, the reconstructed surface deviated by two point six zero millimeters root mean square over the two hundred twenty-five millimeter span.
Dev: That result is specifically for the quasi static regime, and it also suggests that adding heterogeneous nodes carrying force or pressure sensors could be done toward zero moment point estimation and ground reaction force estimation.
Taro: It’s interesting how they tackle the power delivery at three point three volts per node, rather than having a regulator on every single sensor node which would add mass and size <ref:2610.19969#pg1>.
Rosa: Exactly, they showed a trade off between frame rate, node count, and node mass—which bound you hit first depends on the operating point of your system.
Dev: So while this is a neat way to get motion capture outside the lab with no wires or batteries, it's important to remember that long term use would require periodic recalibration or adding a magnetometer for an absolute heading reference.
Taro: That's what I was thinking about—the drift issue is significant if you need high accuracy over time without external correction.
Rosa: It really emphasizes that motion capture in the full sense, tracking dynamic deformation, will still require comparing these results against optical motion capture systems to see how they stack up.
Conclusion: Rosa: So we’ve looked at this system that uses conductive fabric to power and talk all those IMU sensor nodes without any wires or batteries at each point.
Dev: That’s the core of it, Rosa, using copper nickel plated threads as a shared medium for both data and DC power delivery.
Rosa: Right. The authors are looking at how you can build this kind of location independent platform for motion capture that works outside the lab or in real life applications like sports or rehab.
Dev: They use a token passing protocol combined with pipelined processing to keep the data acquisition collision free and stable at fifty Hertz.
Rosa: So, against a cylinder of known radius, they got a reconstruction error of two point six millimeters root mean square over the twenty-two five millimeter span.
Dev: That result is for the quasi static regime, which is what they tested empirically with their setup.
Rosa: But then they admit that because each node uses gravity as its only attitude reference, yaw drift adds up, and the flat state error grew from one point four millimeters to twenty-nine point five millimeters over one minute.
Dev: That’s a big caveat for me on the safety case side; you can't rely on that without some kind of absolute heading reference like a magnetometer.
Rosa: Exactly. So, it shows how you can build a batteryless system, but it’s not ready for full dynamic motion capture yet because of that drift issue.
Dev: It means the next big step for this technology is figuring out how to handle those dynamic deformations and characterizing them against optical motion capture systems.
Ryusei Miyakawa, Akihito Noda
School of Systems Engineering, Kochi University of Technology
eess.SP, cs.NI
Submitted: 2026-10-14
Updated: 2026-10-14
Comments: 4 pages, 4 figures, 1 table. Accepted for publication in IEEE Sensors Letters
Journal ref: IEEE Sensors Letters, vol. 10, no. 11, Art. no. 7005504, Nov. 2026
DOI: 10.1109/LSENS.2026.3744423
License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
Importance score: 76/100
The gist: Motion capture is increasingly important in healthcare, 3D Reconstruction rehabilitation, and sports applications This paper proposes a location-independent platform for motion capture…
Key concepts
- Conductive Fabric Bus
- The fabric is woven with copper–nickel threads, serving as a shared medium for both power delivery and data communication. This eliminates the need for individual wiring or onboard batteries at each sensor node, creating a flexible and wireless network where nodes can share resources directly through the textile.
- Token-Passing Protocol
- This communication method ensures collision-free and stable data acquisition by assigning transmission rights sequentially. Each node broadcasts a token to signal its turn to transmit data, which is crucial for maintaining reliable real-time operation across the sensor network.
- Quaternion Fusion (Madgwick Filter)
- Each IMU node fuses its raw acceleration and angular velocity data into a quaternion using the Madgwick complementary filter running on the MCU. This process converts noisy sensor readings into a stable orientation estimate, which is essential for accurately determining the 3D posture of the sensor.
- Shape Reconstruction via Slerp
- The system reconstructs the fabric's surface by combining IMU orientations with its known grid layout. Adjacent node positions are estimated using fixed-length displacement vectors whose directions are determined by spherical linear interpolation (Slerp) based on intermediate orientations.
Terminology
Summary
Motion capture is increasingly important in healthcare, 3D Reconstruction rehabilitation, and sports applications This paper proposes a location-independent platform for motion capture, based on an array of IMU (inertial measurement unit) sensor nodes without individual wiring or onboard batteries, enabled by two-dimensional communication and power delivery over conductive fabric
System Architecture
The proposed system utilizes conductive fabric as a shared two-dimensional medium for both power delivery and data communication The hardware platform involves IMU sensor nodes placed directly on the fabric, with a master node supplying power while collecting posture data The fabric is woven with copper–nickel (Cu–Ni)-plated threads, serving as a two-dimensional shared medium for power and data
Communication Protocol
A token-passing protocol combined with per-node pipelining enables collision-free and stable data acquisition at 50 Hz Each node fuses the three-axis acceleration and three-axis angular velocity of its IMU into a quaternion using a Madgwick complementary filter on the MCU The protocol assigns transmission rights sequentially, with each node broadcasting a token naming its successor Sensing, computation and communication are pipelined within each node, meaning every transmission is broadcast two slots before its turn
Reconstruction Methodology
The shape of the fabric is reconstructed by combining IMU orientations with the known grid arrangement of the nodes Adjacent node positions are estimated from fixed-length displacement vectors whose directions follow an intermediate orientation computed by spherical linear interpolation (Slerp) The position of node j is estimated from that of node i as p j = p i + q i, j Lv i, j q-1 i, j A continuous surface follows by cubic Hermite interpolation of the coordinates and the tangents derived from the node orientations
Performance Metrics
Against a cylinder of known radius, the reconstructed surface deviates by 2.60 mm RMS over the 225 mm span of the array The reconstruction is formulated for the quasi-static regime, and this letter empirically validates it in that regime The system sustains a 50 Hz frame rate across 16 nodes The mean period was 20. p. 0206 ms with a standard deviation of 0. p. 2127 ms, so the schedule is deterministic to 0. p. 26%
Limitations and Future Work
Four limitations should be noted, including that each node derives orientation from a filter that uses gravity as attitude reference Because the nodes have no absolute heading reference, yaw drift accumulates, and the flat-state error grew from 1.4 mm to 29.5 mm over 1 min The surface is assumed inextensible, and accuracy should degrade on nondevelopable surfaces and under in-plane strain Long-term use would therefore require periodic recalibration, or a magnetometer to supply the missing reference The paper concludes that motion capture in the full sense will require a reconstruction formulated for dynamic deformation and its characterization against optical motion capture
Conclusion
This paper presented a batteryless and wire-free shape reconstruction system using conductive fabric Using the fabric as a shared power and data bus, and superimposing data on DC power via bias-tee circuits, eliminated the need for individual batteries and wiring at each sensor node A token-passing protocol combined with pipelined data processing achieved collision-free real-time data acquisition at 50 Hz in a 4×4 sensor network Against a cylinder of known radius the reconstructed surface deviated by 2.60 mm RMS over the 225 mm span of the array, obtained without a predefined skeletal model The 50 Hz figure is a property of the acquisition system rather than a validated dynamic accuracy The paper also suggests that heterogeneous nodes carrying force or pressure sensors can be added toward zero-moment-point (ZMP) and ground-reaction-force estimation
ACKNOWLEDGMENT
This research was supported in part by JSPS KAKENHI JP25K03109 and JST CRONOS, Japan, Grant Number JPMJCS25N4
REFERENCES
[1] L. Carcreff et al. p. 3, “Validity, reliability and interpretability of an IMU-based system to measure 3D lower limb kinematics of patients with heterogeneous gait disorders,” Sensors, vol. p. 3 26, 2026, art. p. 3 no. 1746.ref:2610.19969V[1]
[2] E. van der Kruk and M. p. 3 M. Reijne, “Accuracy of human motion capture systems for sport applications; state-of-the-art review,” European Journal of Sport Science, vol. p. 3 18, no. 6, pp. 806–819, 2018. p. 3ref:2610.19969V[2]
[3] J. C. M. Schlachetzki et al. p. 3, “Wearable sensors objectively measure gait parameters in Parkinson’s disease,” PLOS ONE, vol. p. 3 12, no. 10, 2017, art. p. 3 no. e0183989.ref:2610.19969V[3]
[4] V. B. Semwal et al. p. 3, “Pattern identification of different human joints for different human walking styles using inertial measurement unit (IMU) sensor,” Artificial Intelligence Review, vol. p. 3 55, pp. 1149–1169, 2022. p. 3ref:2610.19969V[4]
[5] M. Gregorciˇ c and D. p. 3 Georgiev, “The usefulness of wearable sensors for detecting freezing of gait in Parkinson’s disease: A systematic review,” Sensors, vol. p. 3 25, no. 16, 2025, art. p. 3 no. 5101.ref:2610.19969V[5]
[6] Z. K. Lim et al., “Fall risk prediction using temporal gait features and machine learning approaches,” Frontiers in Artificial Intelligence, vol. p. 3 7, 2024, art. p. 3 no. 1425713.ref:2610.19969V[6]
[7] N. Gaud et al. p. 3, “FIBiLS: Fall detection of healthy elderly using IMU sensor and BiLSTM model,” IEEE Sensors Journal, vol. p. 3 25, no. 19, pp. 37 124–37 131, 2025. p. 3ref:2610.19969V[7]
[8] B. Adlou, C. Wilburn, and W. p. 3 Weimar, “Motion capture technologies for athletic performance enhancement and injury risk assessment: A review for multi-sport organizations,” Sensors, vol. p. 3 25, 2025, art. p. 3 no. 4384.ref:2610.19969V[8]
[9] H. Joo, T. Simon, and Y. p. 3 Sheikh, “Total capture: A 3D deformation model for tracking faces, hands, and bodies,” in Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, 2018, pp. p. 3 8320–8329. p. 3ref:2610.19969V[9]
[10] P. Furgale, J. p. 3 Rehder, and R. p. 3 Siegwart, “Unified temporal and spatial calibration for multi-sensor systems,” in 2013 IEEE/RSJ International Conference on Intelligent Robots and Systems, 2013, pp. p. 3 1280–1286. p. 3ref:2610.19969V[10]
[14] S. O. H. Madgwick, “An efficient orientation filter for inertial and inertial/magnetic sensor arrays,” University of Bristol, Tech. p. 3 Rep., 2010.ref:2610.19969V[14]
[15] R. Mahony, T. Hamel, and J. p. 3-M. Pflimlin, “Nonlinear complementary filters on the special orthogonal group,” IEEE Transactions on Automatic Control, vol. p. 3 53, no. 5, pp. 1203–1218, 2008. p. 3ref:2610.19969V[15]
[16] Sony Corporation, “mocopi developer site,” 2023, accessed: 2026-05-04.ref:2610.19969V[16]
[17] F. Grutzmacher ¨ et al. p. 3, “fastSW: Efficient piecewise linear approximation of quaternion-based orientation sensor signals for motion capturing with wearable IMUs,” Sensors, vol. p. 3 21, no. 15, 2021, art. p. 3 no. 5180.ref:2610.19969V[17]
[18] A. Tsukamoto, K. p. 3 Mase, and Y. p. 3 Enokibori, “Integrated utilization of IMU-based human activity recognition datasets across varied configurations using MIG HAR dataset,” International Journal of Activity and Behavior Computing, vol. p. 3 2024, no. 2, pp. 1–21, 2024. p. 3ref:2610.19969V[18]
[19] A. Noda and H. p. 0 Shinoda, “Inter-IC for wearables (I2We): Power and data transfer over double-sided conductive textile,” IEEE Transactions on Biomedical Circuits and Systems, vol. p. 3 13, no. 1, pp. 80–90, 2019. p. 3ref:2610.19969V[19]
[20] A. Noda, “UART for wearables (U4We): DC power and carrierless signal transfer over conductive textiles,” IEEE Transactions on Circuits and Systems—II: Express Briefs, 2026, doi: 10. p. 31109/TCSII. p. 32026.3735640.ref:2610.19969V[20]
[21] M. Magri and P. p. 3 Ciarletta, “Mathematics meets the fashion industry on path to product innovation and sustainability,” Proceedings of the Royal Society A: Mathematical, Physical and Engineering Sciences, vol. p. 3 479, 2023, art. p. 3 no. 20220715.ref:2610.19969V[21]
[23] K. Shoemake, “Animating rotation with quaternion curves,” in Proceedings of the 12th Annual Conference on Computer Graphics and Interactive Techniques, ser. p. 3 SIGGRAPH ’85, vol. p. 3 19, no. 3, 1985, pp. p. 3 245–254.ref:2610.19969V[23]
[24] J. Gonzalez-Alonso ´ et al. p. 3, “Custom IMU-based wearable system for robust 2. p. 34 GHz wireless human body parts orientation tracking and 3D movement visualization on an avatar,” Sensors, vol. p. 3 21, 2021, art. p. 3 no. 6642.ref:2610.19969V[24]
[25] N. Maske et al. p. 3, “ZMP estimation from wearable sensor using deep learning for gait analysis,” IEEE Sensors Letters, vol. p. 3 10, no. 4, pp. 1–4, 2026, art. p. 3 no. 5502004.<ref:2610.
Green — the paper says it, and the page is printed. Red — no passage carries it, so treat that sentence as unverified. The same check runs on your own text: check a summary against its source, $5.
Improvements for AI systems
-
Improve real-time shape reconstruction by integrating dynamic deformation tracking into the estimation pipeline to handle accelerations that rival gravity, moving beyond
quasi-static regime.
This would allow AI systems to predict or track object movement in real-time using IMU data without relying solely on a static reconstruction model. -
Develop a system capable of estimating absolute heading by incorporating a magnetometer into the sensor node design, as the current system has
no absolute heading reference
leading to yaw drift. This would enable AI systems to maintain accurate spatial orientation even during prolonged motion or complex deformations. -
Implement machine learning models that utilize the
token-passing protocol
data stream to perform predictive filtering on IMU inputs within each node, allowing for proactive rather than purely reactive data processing. This would improve the robustness of the 50 Hz frame rate by anticipating noise or transient events before they are fully processed. -
Create a heterogeneous sensor network capable of incorporating force and pressure sensors onto the conductive fabric bus to estimate zero-moment-point (ZMP) and ground-reaction forces. This would allow AI systems to perform advanced gait analysis, fall risk assessment, or dynamic balance control based on actual physical interaction with the environment.
-
Enhance the reconstruction algorithm by explicitly modeling
in-plane stretching
andconstant internode distance
as dynamic variables within a simplified mass-spring model framework rather than assuming them as negligible constants. This would allow AI systems to better characterize the material properties of the textile surface under strain during deformation.
Related papers
- Open-H-RF: A Large-Scale Ultrasound Raw Channel Dataset Initiative
- Scalable Attention for 5G NR Channel Estimation
- Low-Altitude Wireless Networks: A Comprehensive Survey
- Sensing-Assisted OFDM with Index Modulation: A Waveform for Optical Wireless Integrated Communication and Localisation
- Iteration-Free Candidate Pruning and an Index-Domain VLSI Architecture for Sort-Free MIMO Box Decoding up to 1024-QAM
- A Foundation Model for Large-Scale Wireless Network Planning, Operation and Optimization