Data-centric neuromotor interfaces for portable human-machine interaction

Abstract

Dexterous human-machine interaction requires intuitive and expressive interfaces that can be efficiently deployed on constrained edge devices. Flexible material-based neuromotor interfaces hold considerable promise, as they decode human movement intention into natural control. Although emerging flexible electronic skins enable wearable high-fidelity data acquisition, practical deployment inevitably involves trade-offs between computational resources and portability. We present a data-centric paradigm where physiological features yield fundamental separability, providing sufficient discriminative cues for recognition. A wireless, high-bandwidth system developed for collecting various electrophysiological signals, when integrated with muscle-specific electrodes, forms a surface electromyography-based interface. Exploiting highly separable data, a 2,210-parameter model achieves 94.36% accuracy across 34 gestures and can be rapidly deployed on edge devices, establishing a new thousand-parameter benchmark for dexterous decoding. The underlying data–algorithm interactions in the data-centric paradigm are further clarified, demonstrating its feasibility in real-world scenarios. This study provides a principled and validated pathway for practical deployment of reliable neuromotor interfaces.

Keywords

Flexible electrode array / gesture recognition / lightweight algorithm / signal separability / surface electromyography / wearable sensing

References

[1] Zhang Y,Chen B.Force-aware interface via electromyography for natural VR/AR interaction.ACM Trans Graph2022;41:1-18

[2] Hong J,Chen Y.Body-coupled-driven object-oriented natural interactive interface.Adv Mater2025;37:e07067

[3] Liu L,Li S.High-performance, noninvasive flexible sensing array for gesture and handwriting recognition assisted by machine learning.Soft Sci2026;6:34

[4] Gu G,Xu H.A soft neuroprosthetic hand providing simultaneous myoelectric control and tactile feedback.Nat Biomed Eng2023;7:589-98

[5] Capsi-Morales P,Catalano MG,Bicchi A.Merging motoneuron and postural synergies in prosthetic hand design for natural bionic interfacing.Sci Robot2025;10:eado9509

[6] Xie C,Li G.A cable-driven upper limb rehabilitation robot with muscle-synergy-based myoelectric controller.IEEE Trans Robot2024;40:3199-211

[7] Zhong X,Liu D.Plug-and-play sEMG-driven hand gesture recognition with subdomain adaptation for exoskeleton rehabilitation gloves.IEEE Trans Instrum Meas2025;74:1-10

[8] Marcos-Antón S,Oña-Simbaña ED,Martínez-Rolando L.sEMG-controlled forearm bracelet and serious game-based rehabilitation for training manual dexterity in people with multiple sclerosis: a randomised controlled trial.J Neuroeng Rehabil2023;20:110 PMCID:PMC10440030

[9] Gao Y,Zhang C.Rehabilitation exoskeleton system with bidirectional virtual reality feedback training strategy.CAAI Trans Intell Technol2025;10:728-37

[10] Xiao Y,Gao Y.Interactive virtual ankle movement controlled by wrist sEMG improves motor imagery: an exploratory study.IEEE Trans Vis Comput Graph2024;30:5507-24

[11] Yiu C,Park W.Skin-interfaced multimodal sensing and tactile feedback system as enhanced human-machine interface for closed-loop drone control.Sci Adv2025;11:eadt6041 PMCID:PMC11939050

[12] Chen M.High-density surface EMG decomposition: achievements, challenges, and concerns.IEEE Trans Neural Syst Rehabil Eng2025;33:1212-9

[13] Zhao H,Chen M.Online decomposition of surface electromyogram into individual motor unit activities using progressive FastICA peel-off.IEEE Trans Biomed Eng2024;71:160-70

[14] Prabhavathy T,Balaji E.Hand gesture classification framework leveraging the entropy features from sEMG signals and VMD augmented multi-class SVM.Expert Syst Appl2024;238:121972

[15] Liu Y,Yang L,Yu H.A CNN-transformer hybrid recognition approach for sEMG-based dynamic gesture prediction.IEEE Trans Instrum Meas2023;72:1-16

[16] Wang Z,Xu M,Su J.Transformer-based network with temporal depthwise convolutions for sEMG recognition.Pattern Recognit2024;145:109967

[17] Zhang Y,Yu H,Lu W.Learning effective spatial–temporal features for sEMG armband-based gesture recognition.IEEE Internet Things J2020;7:6979-92

[18] Xiong B,Niu Y,Mao G.A global and local feature fused CNN architecture for the sEMG-based hand gesture recognition.Comput Biol Med2023;166:107497

[19] Gan Z,Wu P.SGRN: SEMG-based gesture recognition network with multi-dimensional feature extraction and multi-branch information fusion.Expert Syst Appl2025;259:125302

[20] He D,Wang X,Zhao Y.DSRANet: a lightweight depthwise separable residual attention network for sEMG gesture recognition.Biomed Signal Process Control2026;113:108845

[21] Zhang A,Zou J.Shifting machine learning for healthcare from development to deployment and from models to data.Nat Biomed Eng2022;6:1330-45 PMCID:PMC12063568

[22] Kaifosh P.CTRL-labs at Reality LabsA generic non-invasive neuromotor interface for human-computer interaction.Nature2025;645:702-11 PMCID:PMC12443603

[23] Bai C,Liu Q.A self-powered hydrogel electronic skin with decoupled multimodal sensing for closed-loop human-machine interactions.Nat Commun2026;17:2675 PMCID:PMC13009196

[24] Wang C,Deng Y.Wireless, wrist‐worn ultraconformal neuromuscular interfaces for thumb‐controlled immersive human–machine interactions.Adv Funct Mater2025;35:2422980

[25] Chamberland F,Tam S.Novel wearable HD-EMG sensor with shift-robust gesture recognition using deep learning.IEEE Trans Biomed Circuits Syst2023;17:968-84

[26] Moin A,Rahimi A.A wearable biosensing system with in-sensor adaptive machine learning for hand gesture recognition.Nat Electron2021;4:54-63

[27] Sîmpetru RC,Simon AU.MyoGestic: EMG interfacing framework for decoding multiple spared motor dimensions in individuals with neural lesions.Sci Adv2025;11:eads9150 PMCID:PMC11980827

[28] Wu S,Li K,Li Y.Environmentally friendly high-performance electronic skin for intelligent human–machine interaction systems with feedback capabilities.J Mater Sci Technol2027;276:122-32

[29] Farina D,Enoka RM.The extraction of neural strategies from the surface EMG: 2004-2024.J Appl Physiol2025;138:121-35

[30] Ives JC.Sampling rate effects on surface EMG timing and amplitude measures.Clin Biomech2003;18:543-52

[31] Stegeman DF,Lapatki BG.High-density surface EMG: techniques and applications at a motor unit level.Biocybern Biomed Eng2012;32:3-27

[32] Botros FS,Scheme EJ.Electromyography-based gesture recognition: is it time to change focus from the forearm to the wrist?.IEEE Trans Ind Infrom2022;18:174-84

[33] Pelaez Murciego, L.; Henrich, M. C.; Spaich, E. G.; Dosen, S. Reducing the number of EMG electrodes during online hand gesture classification with changing wrist positions.J Neuroeng Rehabil2022;19:78 PMCID:PMC9306156

[34] Hermens HJ,Disselhorst-Klug C.Development of recommendations for SEMG sensors and sensor placement procedures.J Electromyogr Kinesiol2000;10:361-74

[35] Myo Armband. https://wearabletech.io/myo-bracelet/. (accessed 2026-09-10)

[36] Elizondo DA,Gamez M,Alfaro E.Linear separability and classification complexity.Expert Syst Appl2012;39:7796-807

[37] Davies DL.A cluster separation measure.IEEE Trans Pattern Anal Mach Intell1979;PAMI-1:224-7

[38] Yang S,Shang J.Stretchable surface electromyography electrode array patch for tendon location and muscle injury prevention.Nat Commun2023;14:6494 PMCID:PMC10576757

[39] Du Q,Sun S.Self-compliant ionic nanomesh for gas-permeable and stress-free on-skin electronics.Nat Commun2025;16:11510 PMCID:PMC12749131

[40] Zhu S,Fang B.Progress and prospect for conducting polymer fibers.Adv Mater2025;37:e04071

[41] Zhu Y,Chen B,Liang C.Physical human–robot interaction control of variable stiffness exoskeleton with sEMG-based torque estimation.IEEE Trans Ind Inform2023;19:10601-12

[42] Zhu Y,Chen B.Design and voluntary control of variable stiffness exoskeleton based on sEMG driven model.IEEE Robot Autom Lett2022;7:5787-94

[43] Sierotowicz M,Nell L.EMG-driven machine learning control of a soft glove for grasping assistance and rehabilitation.IEEE Robot Autom Lett2022;7:1566-73

[44] Hannun AY,Haghpanahi M.Cardiologist-level arrhythmia detection and classification in ambulatory electrocardiograms using a deep neural network.Nat Med2019;25:65-9 PMCID:PMC6784839

[45] Tang X,Zhao S,Liu J.Flexible brain–computer interfaces.Nat Electron2023;6:109-18

[46] Qian y,Yu P.Real-time decoding of full-spectrum chinese using brain-computer interface.Sci Adv2025;11:eadz9968 PMCID:PMC12588286

[47] Défossez A,Rapin J,King J.Decoding speech perception from non-invasive brain recordings.Nat Mach Intell2023;5:1097-107

[48] Yang C,Liu C,Dai S.Haptics electromyography perception and learning enhanced intelligence for teleoperated robot.IEEE Trans Autom Sci Eng2019;16:1512-21

[49] Ogenyi UE,Yang C,Liu H.Physical human-robot collaboration: robotic systems, learning methods, collaborative strategies, sensors, and actuators.IEEE Trans Cybern2021;51:1888-901

[50] Capallera M,Meteier Q,Mugellini E.Human-vehicle interaction to support driver’s situation awareness in automated vehicles: a systematic review.IEEE Trans Intell Veh2023;8:2551-67

[51] Shen Z,Hou Y,Yang H.High-resolution flexible tactile sensors.Soft Sci2026;6:53