Pith. sign in

REVIEW 4 major objections 5 minor 65 references

Performance Benchmarking of Psychomotor Skills Using Wearable Devices: An Application in Sport

T0 review · 4 major / 5 minor · reviewed 2026-08-12 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read Wearable sensors can map sports strokes into a performance space and identify the optimal technique by clustering.

desk verdict A coherent IMU-to-performance-space pipeline for table tennis forehand benchmarking, but the 'benchmark cluster' label is an artifact of unreported expert-chosen normalizations and is never validated against skill or outcomes. read the letter →

arxiv 2411.16168 v2 pith:U6PZACTS submitted 2024-11-25 eess.SP

Mahela Pandukabhaya , Tharaka Fonseka , Madhumini Kulathunge , Roshan Godaliyadda , Parakrama Ekanayake , Chanaka Senanayake , Vijitha Herath This is my paper · ORCID
classification eess.SP
keywords wearableIMUsensorspsychomotorskillbenchmarkingperformancespacebenchmarkclusterspectralclusteringk-meansEuleranglestabletennisforehand
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

The paper claims a general framework can benchmark psychomotor skills by converting wearable motion data into a performance space and clustering it to identify a benchmark of optimal executions. It validates the idea on the table tennis forehand stroke with 18 players of mixed skill, each performing 50 forehands while wearing four IMU sensors on the dominant arm; every stroke becomes a point in a five-dimensional space whose coordinates are expert-defined quality metrics (bounce distance X and Y, net clearance, instantaneous ball speed, height ratio), normalized to scores in $[0,1]$. Spectral and k-means clustering of that space produced four clusters, and the cluster closest to the ideal point $(1,1,1,1,1)$ was declared the benchmark; differences between clusters are read back as Euler-angle deviations at wrist, elbow, and shoulder. If this works as described, it offers a low-cost, data-driven route to individualized coaching and skill assessment in sports, rehabilitation, and industrial training.

What carries the argument

The central object is the performance space $\mathcal{S} \subset [0,1]^{n_p}$, built from $n_p=5$ expert-chosen quality parameters. Each parameter passes through a cost function and then a Gaussian radial-basis function $\psi_p(x_p) = \psi_0 e^{-\alpha_p^2 (x_p-\mu_p)^2}$, which inverts the cost into a profit score in $[0,\psi_0]$; the ideal point $O'=(1,\ldots,1)$ is unattainable but serves as the reference for benchmarking. The cluster with the smallest Euclidean distance from its centroid to $O'$ is declared the benchmark. The other load-bearing pieces are the extended Kalman filter that fuses raw accelerometer and gyroscope data into quaternions and then Euler angles (a reduction the paper calls a contraction mapping), and the $\sigma$-sweep spectral-clustering step that fixes the number of clusters before k-means is applied. The final correspondence -- reading performance-space deviations back as Euler-angle differences at wrist, elbow, and shoulder -- is the mechanism that converts a statistical cluster into a kinematic explanation and hence into coaching advice.

What would settle it

Have coaches or match outcomes independently rate the same forehand strokes without seeing the cluster labels, and compare those ratings with cluster membership. If strokes in the deviating clusters are rated superior or win points more often than benchmark-cluster strokes, the framework's notion of optimal performance fails; the simplest version is a blind study in which expert coaches rank a random sample of strokes that the framework labeled benchmark versus deviating.

Watch

Extended reading notes

Core claim

The paper's central claim is that raw IMU motion data, after being reduced from a 24-dimensional signal space to a 9-dimensional Euler-angle space by an extended Kalman filter, can be mapped into a performance space in which unsupervised clustering reveals a benchmark cluster of optimal strokes and a small number of deviating clusters. For the table-tennis case study, the mapping uses five quality parameters -- bounce distances $X$ and $Y$, net clearance, instantaneous ball speed, and height ratio -- passed through expert-chosen cost functions and Gaussian radial-basis normalizations so that each stroke receives a profit score in $[0,1]^5$. The benchmark cluster is defined as the one whose centroid is nearest the ideal (unattainable) point $O'=(1,1,1,1,1)$; here that is cluster 0, with Euclidean distance $d_0 = 0.6284$, and it contains strokes from all 18 players, showing that the benchmark captures good executions rather than a particular player's identity. The paper then compares mean Euler-angle signals across clusters and finds that the benchmark strokes are characterized by steady wrist yaw, a subtle forward wrist pitch, and a smooth elbow push followed by retraction, whereas deviating clusters show larger wrist fluctuations and reversed elbow/shoulder patterns. These correspondences are claimed to be a breakdown analysis that maps performance-space distance back to concrete kinematic corrections.

Load-bearing premise

The assumption that carries the whole framework is that the five quality metrics plus the expert-chosen scoring curves truly capture what makes a forehand stroke effective; if a stroke that wins points scores poorly on bounce distances, net clearance, speed, or height ratio, the 'benchmark' cluster simply reflects the scoring choices, not the skill being measured.

Editorial extensions

If this is right

  • Any psychomotor task with measurable quality indicators -- surgical gestures, industrial assembly motions, rehabilitation exercises -- could be benchmarked with the same pipeline by swapping in task-specific parameters.
  • Because benchmark-cluster strokes come from every player in the study, the method separates good and bad executions within each individual rather than merely ranking players by experience level.
  • Euler-angle differences between benchmark and deviating clusters give coaches concrete kinematic targets, such as reducing wrist roll fluctuation or reversing the elbow rotation pattern.
  • Sub-clustering a deviating cluster can expose finer distinctions, for example a wrist-control deficit that splits one flawed group into two, allowing more targeted interventions.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • Extension not tested in the paper: the benchmark definition is validated only internally, so a blind external rating study (coaches or match outcomes) would test whether the cluster truly tracks skill rather than the five chosen metrics.
  • The framework assumes deep, wide, fast, low-clearance strokes are always better, but tactical play sometimes favors short pushes; testing strokes with varied tactical intent would reveal whether the benchmark is a universal skill measure or a narrow tactical preference.
  • The number of clusters (four) is selected by reading which spectral eigengap dominates the sigma sweep; a stability analysis across many sigma values would tell whether the four-cluster structure is robust or an artifact of the sweep.
  • Swapping the five table-tennis parameters for clinical metrics (range of motion, smoothness, symmetry) would test the framework's generality in rehabilitation, where the 'optimal' target is defined differently than in sport.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

4 major / 5 minor

Summary. The paper proposes a framework for benchmarking psychomotor skills using a wearable IMU system. Motion trajectories are mapped into a performance space via five table-tennis-specific quality metrics (bounce distances X and Y, net clearance, instantaneous ball speed, height ratio) using expert-defined cost functions and Gaussian RBF normalization; the resulting [0,1]^5 space is clustered with spectral clustering (sigma sweep) and k-means, and the benchmark cluster is defined as the one whose centroid lies closest to the ideal point O' = (1, ..., 1). The authors then compute mean Euler-angle signals per cluster to interpret deviations and demonstrate the pipeline on forehand strokes from 18 players. The manuscript claims that this framework identifies optimal performance and provides a basis for coaching feedback.

Significance. If the central claim were established, the framework would be a useful low-cost approach to psychomotor skill benchmarking, with plausible transfer to rehabilitation and vocational training. The paper's strengths are its end-to-end instantiation: ethical clearance, hardware description, EKF-based Euler-angle estimation, a transparent clustering pipeline, and a concrete case study with t-SNE visualizations and per-cluster mean Euler signals. However, the claim that the selected cluster represents 'optimal performance' is not supported by the evidence: the benchmark is defined relative to an expert-chosen ideal point, the cluster count is chosen by a subjective eigengap rule, and no external validation against skill level, coach ratings, or match outcomes is provided. The significance is therefore conditional on either additional validation or a careful reframing of the claims.

major comments (4)
  1. [Section II-G, Eq. (21)] The benchmark cluster is selected as the argmin of centroid distance to O' = (1, ..., 1), an a priori ideal point; consequently, the pipeline always outputs some cluster as 'benchmark' even if no stroke in the sample is actually good, and the label 'optimal' is a property of the expert-chosen cost functions and RBF normalization rather than an independent discovery. Section IV-D reports that the benchmark cluster contains strokes from all 18 players, but no statistical test is performed to check whether membership is associated with the expertise levels in Table 3, and no comparison with coach ratings or match outcomes is offered. The manuscript should either validate the benchmark against an external anchor or explicitly reframe the claims as descriptive of the chosen quality criteria.
  2. [Section IV-B, Fig. 8] The choice Ksc = 4 is based on the eigengap curve after 'excluding the first two eigengaps, which are trivial cases,' but no objective criterion, σ-range, or resolution for the sigma sweep is given, and no stability analysis or bootstrapping is reported. Since the cluster count is load-bearing for all subsequent results, the paper needs either a reproducible rule for selecting Ksc or a sensitivity analysis showing that the benchmark selection is robust to this choice.
  3. [Section II-E, Eq. (15)] The Gaussian RBF shape parameters α_p and centers μ_p are expert choices that are never reported, although they determine the geometry of the performance space and can reorder the centroid distances in Table 5, thereby changing which cluster is selected as benchmark. The parameter values and the derivation of the cost functions in Fig. 7a should be provided, and a robustness check over plausible parameter ranges should be reported.
  4. [Sections IV-F and IV-G] The Euler-angle interpretations are produced after the benchmark cluster has been chosen, so they cannot independently confirm that the selected cluster is optimal; moreover, the comparisons are qualitative (for example, 'exact opposite movement') with no confidence intervals, hypothesis tests, or variance information on the mean signals. The absence of statistical evidence is especially relevant because the number of usable realizations per cluster is not stated, and the t-SNE visualizations in Figs. 9 and 13 are not accompanied by quantitative cluster-quality measures.
minor comments (5)
  1. [Equations (9)-(13) and (20)] Several equations contain typesetting artifacts (for example, '1ˆxk', 'δ ˆθk', and '/radicaltp/radicalvertex/radicalvertex√'), which should be corrected before publication.
  2. [Abstract] The phrase 'thorough validation' in the abstract is stronger than what is actually presented; suggest softening to 'demonstration' or adding direct validation evidence.
  3. [Section II-F] The sigma-sweep procedure is described verbally as 'modes of clustering' but without a formal algorithm or a precise rule for selecting Ksc; a numbered algorithmic description would improve reproducibility.
  4. [Section IV-A and Fig. 7] The cost functions in Fig. 7a are shown graphically but their analytic forms are not given; this makes it impossible to reproduce the RBF mapping or assess the effect of expert choices.
  5. [General] No data or code availability statement is provided; given that the RBF parameters are not reported, a data/code availability statement would materially aid evaluation and replication.

Circularity Check

1 steps flagged · score 6.0 of 10

The benchmark cluster is chosen as the cluster closest to the expert-constructed ideal point in the RBF-defined performance space; its “optimal” label restates the selection criterion rather than deriving an independent result.

  1. self definitional [Section II-E2 (Eq. 15), Section II-G (Eqs. 19-21), and Section V conclusion]
    "ψp : R → [0, ψ0] , ψp(xp) = ψ0e−α2 p (xp−µp)2 , (15) where αp ∈ R+ is a shape parameter defined for each parameter p1, . . . ,pnp based on the recommendations by field experts ... We define this point as the origin of ideal performance O′ ≡ (1, . . . ,1) ... Finally, the benchmark cluster is defined as the cluster Cb where: b = argmin q=0,...,Ksc−1(dq), (21) thereby choosing the cluster closest to O′."

    The performance-space coordinates are not derived from the raw motion data alone; they are the outputs of expert-selected Gaussian functions (Eq. 15) whose centers µp and widths αp are set according to expert recommendations. The ideal point O′ = (1,...,1) is by construction the point of maximal expert-defined scores. Eq. (21) then defines the benchmark as the cluster with the smallest Euclidean distance to O′. Thus the conclusion in Section V that this cluster “signifies optimal performance” is logically equivalent to saying it is the nearest cluster to a manually constructed scoring point.

full rationale

The paper builds a genuinely self-contained pipeline from IMU data to Euler angles (EKF) and from video-based measurements to five task parameters. The clustering (spectral + k-means) is unsupervised and operates on the derived performance scores. The motion-space deviation analysis compares cluster mean Euler signals after cluster selection. However, the crucial step of identifying which cluster is the “benchmark” is not data-driven: Eq. (21) picks the cluster whose centroid is nearest to O′ = (1,...,1), a point defined as the maximum of expert-chosen RBF-normalized cost functions (Eq. 15). Calling that cluster “optimal” (Section V) is therefore a restatement of the selection criterion rather than a verified discovery. The paper itself acknowledges the expert dependence (Sections II-E1, III-B) and never validates the selected cluster against any external measure of skill. The circularity is partial: the framework itself is coherent and the motion-space interpretations are post hoc, but the central claim that the chosen cluster represents optimal performance reduces, by construction, to the authors’ choice of quality parameters and RBF ideal point.

Assumptions & free parameters 3 free parameters · 5 assumptions · 0 invented entities

The central result depends on a set of expert-chosen scoring functions and the subjective selection of the number of clusters. These choices are not fitted to outcome data and are not justified quantitatively, so the benchmark cluster's interpretation is conditional on them.

free parameters (3)
  • RBF shape parameters alpha_p and centers mu_p for five quality metrics = not reported, selected by field experts
    Eq. (15) maps each cost to [0, psi0] using Gaussian RBFs; alpha_p and mu_p determine how parameter values translate into performance scores, and they are chosen by expert recommendation (Section II-E2). The resulting scores define which strokes are 'good', so the benchmark cluster depends on them.
  • Cost function definitions J_p(x) = expert-defined piecewise functions (Fig. 7a)
    The cost functions assign a quality cost to each raw parameter before RBF normalization. They are based on domain expertise and literature (Section IV-A1), not on outcome data.
  • Spectral clustering sigma sweep selection of Ksc = Ksc=4 chosen from eigengap curves after excluding first two eigengaps
    The number of clusters is determined by subjective inspection of eigengap versus log(sigma) curves, with the first two gaps excluded as 'trivial'. This choice affects the k-means clustering and hence the benchmark cluster identity.
assumptions (5)
  • domain assumption The five chosen parameters fully characterize optimal table tennis forehand performance.
    Section III-B defines the parameters and asserts that a successful stroke maximizes bounce distances, minimizes net clearance with sufficient margin, achieves high speed, and contacts the ball near maximum height. The benchmark cluster is defined relative to these metrics; if they are incomplete or wrong, the benchmark is not meaningful.
  • ad hoc to paper Expert-defined cost functions and RBF mappings preserve quality ordering, so higher normalized scores correspond to better strokes.
    Section II-E and IV-A construct the performance space S from these functions. There is no empirical validation that the normalized scores agree with coach ratings or match outcomes.
  • domain assumption The spectral clustering eigengap heuristic correctly identifies the number of true clusters in the performance space.
    Section IV-B1 selects Ksc=4 based on the fourth eigengap dominating the sigma sweep after discarding two gaps. This heuristic is standard but subjective and not statistically validated here.
  • domain assumption The EKF-based joint angle estimation from IMU data is accurate enough for the qualitative comparisons.
    Section II-C outlines the EKF. No accuracy validation for this specific sensor setup is included, though the device is said to be validated using YOLOv7 (Section II-B2), which is an object-detection tool, not a motion-capture reference.
  • domain assumption Video-derived parameters (bounce locations, net clearance, speed, height ratio) are accurate.
    Parameters are extracted manually from 60 fps camera feeds using cross-ratio perspective correction (Section III-D). No error analysis or inter-rater reliability is reported.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Performance Benchmarking of Psychomotor Skills Using Wearable Devices: An Application in Sport." pith.science (2026). https://pith.science/paper/U6PZACTS

@misc{pith2026241116168,
  author       = {Pith},
  title        = {Pith review of: Performance Benchmarking of Psychomotor Skills Using Wearable Devices: An Application in Sport},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/U6PZACTS}},
  note         = {Machine review of arXiv:2411.16168}
}
read the original abstract

Mastering psychomotor skills, such as those essential in sports, rehabilitation, and professional training, often requires a precise understanding of motion patterns and performance metrics. This study proposes a versatile framework for optimizing psychomotor learning through human motion analysis. Utilizing a wearable IMU sensor system, the motion trajectories of a given psychomotor task are acquired and then linked to points in a performance space using a predefined set of quality metrics specific to the psychomotor skill. This enables the identification of a benchmark cluster in the performance space, which represents a group of reference points that define optimal performance across multiple criteria, allowing correspondences to be established between the performance clusters and sets of trajectories in the motion space. As a result, common or specific deviations in the performance space can be identified, enabling remedial actions in the motion space to optimize performance. A thorough validation of the proposed framework is done in this paper using a Table Tennis forehand stroke as a case study. The resulting quantitative and visual representation of performance empowers individuals to optimize their skills and achieve peak performance.

Figures

Figures reproduced from arXiv: 2411.16168 by the authors.

Figure 1
Figure 1. FIGURE 1 [PITH_FULL_IMAGE:figures/full_fig_p004_1.png] view at source ↗
Figure 2
Figure 2. FIGURE 2 [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗
Figure 4
Figure 4. FIGURE 4 [PITH_FULL_IMAGE:figures/full_fig_p007_4.png] view at source ↗
Figures from the paper (8 more)
Figure 5
Figure 5. Figure 5: FIGURE 5 [PITH_FULL_IMAGE:figures/full_fig_p008_5.png]
Figure 6
Figure 6. Figure 6: FIGURE 6 [PITH_FULL_IMAGE:figures/full_fig_p008_6.png]
Figure 7
Figure 7. Figure 7: FIGURE 7 [PITH_FULL_IMAGE:figures/full_fig_p009_7.png]
Figure 8
Figure 8. Figure 8: FIGURE 8 [PITH_FULL_IMAGE:figures/full_fig_p009_8.png]
Figure 10
Figure 10. Figure 10: exhibits the cluster composition for each individual. It can be identified that the benchmark cluster includes data points from all individuals, emphasizing that it captures the well-performed forehand strokes of the players irrespective of their level of expertise. F…
Figure 11
Figure 11. Figure 11: FIGURE 11 [PITH_FULL_IMAGE:figures/full_fig_p010_11.png]
Figure 13
Figure 13. Figure 13: FIGURE 13 [PITH_FULL_IMAGE:figures/full_fig_p011_13.png]
Figure 14
Figure 14. Figure 14: FIGURE 14 [PITH_FULL_IMAGE:figures/full_fig_p011_14.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

65 extracted references · 62 canonical work pages

  1. [1]

    J. K. Aggarwal and Q. Cai, ‘‘Human motion analysis: A review,’’ Comput. Vis. Image Understand., vol. 73, no. 3, pp. 428–440, Mar. 1999

    work page 1999

  2. [2]

    L. Wang, W. Hu, and T. Tan, ‘‘Recent developments in human motion analysis,’’ Pattern Recognit., vol. 36, no. 3, pp. 585–601, Mar. 2003

    work page 2003

  3. [3]

    L. Wang, Z. Zhou, and Q. Zou, ‘‘Analysis system for table tennis techniques and tactics using data mining,’’ Soft Comput., vol. 27, no. 19, pp. 14269–14284, Oct. 2023

    work page 2023

  4. [4]

    Poppe, ‘‘Vision-based human motion analysis: An overview,’’ Comput

    R. Poppe, ‘‘Vision-based human motion analysis: An overview,’’ Comput. Vis. Image Understand., vol. 108, nos. 1–2, pp. 4–18, Oct. 2007

    work page 2007

  5. [5]

    G. M. Migliaccio, J. Padulo, and L. Russo, ‘‘The impact of wearable technologies on marginal gains in sports performance: An integrative overview on advances in sports, exercise, and health,’’ Appl. Sci., vol. 14, no. 15, p. 6649, Jul. 2024

    work page 2024

  6. [6]

    Zhang, H

    S. Zhang, H. Zhou, R. Tchantchane, and G. Alici, ‘‘A wearable human– machine-interface (HMI) system based on colocated EMG-pFMG sensing for hand gesture recognition,’’ IEEE/ASME Trans. Mechatronics, early access, Apr. 26, 2024, doi: 10.1109/TMECH.2024.3386929

    arXiv 2024

  7. [7]

    Wei, Z.-Q

    Z. Wei, Z.-Q. Zhang, and S. Q. Xie, ‘‘Continuous motion intention prediction using sEMG for upper-limb rehabilitation: A systematic review of model-based and model-free approaches,’’ IEEE Trans. Neural Syst. Rehabil. Eng., vol. 32, pp. 1487–1504, 2024

    work page 2024

  8. [8]

    Z. Lu, Y . Zhou, L. Hu, J. Zhu, S. Liu, Q. Huang, and Y . Li, ‘‘A wearable human–machine interactive instrument for controlling a wheelchair robotic arm system,’’ IEEE Trans. Instrum. Meas., vol. 73, pp. 1–15, 2024

    work page 2024

Show all 65 references
  1. [9]

    I. H. López-Nava and A. Muñoz-Meléndez, ‘‘Wearable inertial sensors for human motion analysis: A review,’’ IEEE Sensors J., vol. 16, no. 22, pp. 7821–7834, Nov. 2016

    2016

  2. [10]

    F. Su, M. Chen, M. Zhang, X. Wang, L. Wu, C. Liu, J. Wang, S. Chang, and T. Xin, ‘‘Real-time detection of Parkinson’s states based on wearable sensor with embedded machine learning,’’IEEE Sensors J., vol. 24, no. 23, pp. 39101–39110, Dec. 2024

    2024

  3. [11]

    Muro-De-La-Herran, B

    A. Muro-De-La-Herran, B. Garcia-Zapirain, and A. Mendez-Zorrilla, ‘‘Gait analysis methods: An overview of wearable and non-wearable systems, highlighting clinical applications,’’ Sensors, vol. 14, no. 2, pp. 3362–3394, Feb. 2014

    2014

  4. [12]

    Raman, M

    R. Raman, M. Kaul, R. Meenakshi, S. Jayaprakash, D. R. Devi, and S. Murugan, ‘‘IoT applications in sports and fitness: Enhancing performance monitoring and training,’’ in Proc. 2nd Int. Conf. Smart Technol. Smart Nation (SmartTechCon), Aug. 2023, pp. 137–141

    2023

  5. [13]

    Stefana, F

    E. Stefana, F. Marciano, D. Rossi, P . Cocca, and G. Tomasoni, ‘‘Wearable devices for ergonomics: A systematic literature review,’’ Sensors, vol. 21, no. 3, p. 777, Jan. 2021

    2021

  6. [14]

    J. A. Brown, K. S. Whittaker, S. A. Hawkins, T. J. Santo, and D. S. Teyhen, ‘‘Using fitness trackers to increase physical activity in a U.S. army health initiative,’’Health Educ. J., vol. 80, no. 8, pp. 883–896, Dec. 2021

    2021

  7. [15]

    H. Hu, Z. Cao, X. Y ang, H. Xiong, and Y . Lou, ‘‘Performance evaluation of optical motion capture sensors for assembly motion capturing,’’ IEEE Access, vol. 9, pp. 61444–61454, 2021

    2021

  8. [16]

    Thomaz, A

    E. Thomaz, A. Parnami, J. Bidwell, I. Essa, and G. D. Abowd, ‘‘Technolog- ical approaches for addressing privacy concerns when recognizing eating behaviors with wearable cameras,’’ inProc. ACM Int. Joint Conf. Pervasive Ubiquitous Comput., New Y ork, NY , USA, Sep. 2013, pp. 739–748

    2013

  9. [17]

    Lachner, T

    C. Lachner, T. Rausch, and S. Dustdar, ‘‘A privacy preserving system for AI-assisted video analytics,’’ in Proc. IEEE 5th Int. Conf. F og Edge Comput. (ICFEC), May 2021, pp. 74–78

    2021

  10. [18]

    S. C. Mukhopadhyay, ‘‘Wearable sensors for human activity monitoring: A review,’’IEEE Sensors J., vol. 15, no. 3, pp. 1321–1330, Mar. 2015

    2015

  11. [19]

    Heikenfeld, A

    J. Heikenfeld, A. Jajack, J. Rogers, P . Gutruf, L. Tian, T. Pan, R. Li, M. Khine, J. Kim, J. Wang, and J. Kim, ‘‘Wearable sensors: Modalities, challenges, and prospects,’’ Lab Chip, vol. 18, no. 2, pp. 217–248, 2018

    2018

  12. [20]

    A. B. Nigusse, D. A. Mengistie, B. Malengier, G. B. Tseghai, and L. V . Langenhove, ‘‘Wearable smart textiles for long-term electrocar- diography monitoring—A review,’’ Sensors, vol. 21, no. 12, p. 4174, Jun. 2021

    2021

  13. [21]

    Ahmad, R

    N. Ahmad, R. A. R. Ghazilla, N. M. Khairi, and V . Kasi, ‘‘Reviews on various inertial measurement unit (IMU) sensor applications,’’ Int. J. Signal Process. Syst., vol. 1, pp. 256–262, Mar. 2013

    2013

  14. [22]

    Liang, C

    C. Liang, C. Y u, Y . Qin, Y . Wang, and Y . Shi, ‘‘DualRing: Enabling subtle and expressive hand interaction with dual IMU rings,’’ Proc. ACM Interact., Mobile, Wearable Ubiquitous Technol., vol. 5, no. 3, pp. 1–27, Sep. 2021, doi: 10.1145/3478114

    2021 doi

  15. [23]

    Manupibul, R

    U. Manupibul, R. Tanthuwapathom, W. Jarumethitanont, P . Kaimuk, W. Limroongreungrat, and W. Charoensuk, ‘‘Integration of force and IMU sensors for developing low-cost portable gait measurement system in lower extremities,’’Sci. Rep., vol. 13, no. 1, p. 10653, Jun. 2023. VOLUM...

    2023

  16. [24]

    Xu, Z.-X

    P .-F. Xu, Z.-X. Liu, F. Li, and H.-P . Wang, ‘‘A low-cost wearable hand gesture detecting system based on IMU and convolutional neural network,’’ in Proc. 43rd Annu. Int. Conf. IEEE Eng. Med. Biol. Soc. (EMBC), Nov. 2021, pp. 6999–7002

    2021

  17. [25]

    N. D. Kahanowich and A. Sintov, ‘‘Learning human-arm reaching motion using a wearable device in human–robot collaboration,’’ IEEE Access, vol. 12, pp. 24855–24865, 2024

    2024

  18. [26]

    O. C. Santos, ‘‘Artificial intelligence in psychomotor learning: Modeling human motion from inertial sensor data,’’ Int. J. Artif. Intell. Tools, vol. 28, no. 4, Jun. 2019, Art. no. 1940006

    2019

  19. [27]

    Xu and F

    X. Xu and F. Ke, ‘‘From psychomotor to ‘motorpsycho’: Learning through gestures with body sensory technologies,’’ Educ. Technol. Res. Develop., vol. 62, no. 6, pp. 711–741, Dec. 2014

    2014

  20. [28]

    Oropesa, P

    I. Oropesa, P . Sánchez-González, M. K. Chmarra, P . Lamata, R. Pérez-Rodríguez, F. W. Jansen, J. Dankelman, and E. J. Gómez, ‘‘Supervised classification of psychomotor competence in minimally invasive surgery based on instruments motion analysis,’’ Surgical Endoscopy, vol. 28...

    2014

  21. [29]

    H. J. T. Costa, C. Abelairas-Gomez, V . Arufe-Giráldez, J. M. Pazos-Couto, and R. Barcala-Furelos, ‘‘Influence of a physical education plan on psychomotor development profiles of preschool children,’’J. Human Sport Exercise, vol. 10, no. 1, pp. 126–140, 2015

    2015

  22. [30]

    M. H. Fogtmann, K. Grønbæk, and M. K. Ludvigsen, ‘‘Interaction technology for collective and psychomotor training in sports,’’ in Proc. 8th Int. Conf. Adv. Comput. Entertainment Technol., Nov. 2011, pp. 1–8

    2011

  23. [31]

    J. Zhao, J. Guo, Y . Chen, W. Li, P . Zhou, G. Zhu, P . Han, and D. Xu, ‘‘Improving rehabilitation motivation and motor learning ability of stroke patients using different reward strategies: Study protocol for a single- center, randomized controlled trial,’’Frontiers Neurol., ...

    2024

  24. [32]

    Pandukabhaya, T

    M. Pandukabhaya, T. Fonseka, M. Kulathunge, R. Godaliyadda, P . Ekanayake, C. Senanayake, P . Gamage, and V . Herath, ‘‘Quality assessment of welding using regression analysis of biomechanical data,’’ in Proc. Moratuwa Eng. Res. Conf. (MERCon), Aug. 2024, pp. 376–381

    2024

  25. [33]

    Siouli, S

    S. Siouli, S. Makris, E. Romanopoulou, and P . D. Bamidis, ‘‘Cognitive computer training in children with cognitive and learning disabilities: Two interesting case studies,’’ in Proc. 2nd Int. Conf. Technol. Innov. Sports, Health Wellbeing (TISHW), Jun. 2018, pp. 1–6

    2018

  26. [34]

    Kim, J.-H

    D.-J. Kim, J.-H. Kim, W.-Y . So, and E. Choi, ‘‘The effects of a psychomotor training program on physical coordination in children with development delay,’’Iranian J. Public Health, vol. 46, no. 6, pp. 860–862, Jun. 2017

    2017

  27. [35]

    C. Yi, S. Rho, B. Wei, C. Y ang, Z. Ding, Z. Chen, and F. Jiang, ‘‘Detecting and correcting IMU movements during joint angle estimation,’’ IEEE Trans. Instrum. Meas., vol. 71, pp. 1–14, 2022

    2022

  28. [36]

    Pandukabhaya, M

    M. Pandukabhaya, M. Kulathunge, T. Fonseka, R. Godaliyadda, P . Ekanayake, and V . Herath, ‘‘Estimation of angle of significance in human joint angles using extended Kalman filter and principal component analysis,’’ in Proc. 2025 3rd Int. Conf. Appl. Sens. (APSCON), 2025

    2025

  29. [37]

    C.-Y . Wang, A. Bochkovskiy, and H.-Y . M. Liao, ‘‘YOLOv7: Trainable bag-of-freebies sets new state-of-the-art for real-time object detectors,’’ 2022, arXiv:2207.02696

    2022 arXiv

  30. [38]

    M. V . Örnhag, P . Persson, M. Wadenbäck, K. Åström, and A. Heyden, ‘‘Trust your IMU: Consequences of ignoring the IMU drift,’’ in Proc. IEEE/CVF Conf. Comput. Vis. Pattern Recognit. Workshops (CVPRW), Jun. 2022, pp. 4467–4476

    2022

  31. [39]

    Catelani, L

    M. Catelani, L. Ciani, G. Patrizi, R. Singuaroli, M. Carratù, P . Sommella, and A. Pietrosanto, ‘‘Analysis of noise contributions in low-cost IMUs through Allan’s variance,’’ in Proc. IEEE 10th Int. Workshop Metrology Aerosp. (MetroAeroSpace), Jun. 2023, pp. 258–262

    2023

  32. [40]

    E. J. Lefferts, F. L. Markley, and M. D. Shuster, ‘‘Kalman filtering for spacecraft attitude estimation,’’ J. Guid., Control, Dyn., vol. 5, no. 5, pp. 417–429, Sep. 1982

    1982

  33. [41]

    M. M. Azarbeik, H. Razavi, K. Merat, and H. Salarieh, ‘‘Augmenting inertial motion capture with SLAM using EKF and SRUKF data fusion algorithms,’’ Measurement, vol. 222, Nov. 2023, Art. no. 113690

    2023

  34. [42]

    Y . Chen, X. Li, J. Liu, G. Xu, and Z. Ying, ‘‘Exploratory item classification via spectral graph clustering,’’ Appl. Psychol. Meas., vol. 41, no. 8, pp. 579–599, Nov. 2017

    2017

  35. [43]

    Hamad and P

    D. Hamad and P . Biela, ‘‘Introduction to spectral clustering,’’ in Proc. 3rd Int. Conf. Inf. Commun. Technol., Theory Appl., Apr. 2008, pp. 1–6

    2008

  36. [44]

    Thilakasiri, N

    I. Thilakasiri, N. Senarath, W. Dedunupitiya, G. Ilangarathna, B. Samarakoon, L. Ramanayake, R. Godaliyadda, P . Ekanayake, V . Herath, M. Pinnawala, S. Y atigammana, J. Ekanayake, S. Dharmarathne, and G. Tilakaratne, ‘‘Family income does not necessarily lead to a holistic e-l...

    2023 doi

  37. [45]

    R. A. A. Rupasinghe, S. G. M. P . Senanayake, D. A. Padmasiri, M. P . B. Ekanayake, G. M. R. I. Godaliyadda, and J. V . Wijayakulasooriya, ‘‘Modes of clustering for motion pattern analysis in video surveillance,’’ in Proc. IEEE Int. Conf. Inf. Autom. Sustainability (ICIAfS), D...

    2016

  38. [46]

    X. Qin, W. Dai, P . Jiao, W. Wang, and N. Y uan, ‘‘A multi-similarity spectral clustering method for community detection in dynamic networks,’’ Sci. Rep., vol. 6, no. 1, p. 31454, Aug. 2016

    2016

  39. [47]

    van Dam, M

    A. van Dam, M. Dekker, I. Morales-Castilla, M. Á. Rodríguez, D. Wichmann, and M. Baudena, ‘‘Correspondence analysis, spectral clustering and graph embedding: Applications to ecology and economic complexity,’’Sci. Rep., vol. 11, no. 1, p. 8926, Apr. 2021

    2021

  40. [48]

    Lloyd, ‘‘Least squares quantization in PCM,’’ IEEE Trans

    S. Lloyd, ‘‘Least squares quantization in PCM,’’ IEEE Trans. Inf. Theory, vol. IT-28, no. 2, pp. 129–137, Mar. 1982

    1982

  41. [49]

    J. B. MacQueen, ‘‘Some methods for classification and analysis of multivariate observations,’’ in Proc. 5th Berkeley Symp. Math. Statist. Probab., Jan. 1967, pp. 281–297

    1967

  42. [50]

    Kanungo, D

    T. Kanungo, D. M. Mount, N. S. Netanyahu, C. D. Piatko, R. Silverman, and A. Y . Wu, ‘‘An efficient k-means clustering algorithm: Analysis and implementation,’’ IEEE Trans. Pattern Anal. Mach. Intell., vol. 24, no. 7, pp. 881–892, Jul. 2002

    2002

  43. [51]

    K. P . Sinaga and M.-S. Y ang, ‘‘Unsupervised k-means clustering algorithm,’’ IEEE Access, vol. 8, pp. 80716–80727, 2020

    2020

  44. [52]

    P . K. Hughes, N. L. Bhundell, and J. M. Waken, ‘‘Visual and psychomotor performance of elite, intermediate and novice table tennis competitors,’’ Clin. Experim. Optometry, vol. 76, no. 2, pp. 51–60, Mar. 1993

    1993

  45. [53]

    Bańkosz, S

    Z. Bańkosz, S. Winiarski, and I. M. Lanzoni, ‘‘Gender differences in kinematic parameters of topspin forehand and backhand in table tennis,’’ Int. J. Environ. Res. Public Health, vol. 17, no. 16, p. 5742, Aug. 2020

    2020

  46. [54]

    Y . He, X. Lyu, D. Sun, J. S. Baker, and Y . Gu, ‘‘The kinematic analysis of the lower limb during topspin forehand loop between different level table tennis athletes,’’ PeerJ, vol. 9, Mar. 2021, Art. no. e10841

    2021

  47. [55]

    R. J. Bootsma and P . C. V an Wieringen, ‘‘Timing an attacking forehand drive in table tennis,’’ J. Experim. Psychol., Hum. Perception Perform., vol. 16, no. 1, pp. 21–29, Feb. 1990

    1990

  48. [56]

    Iino and T

    Y . Iino and T. Kojima, ‘‘Kinematics of table tennis topspin forehands: Effects of performance level and ball spin,’’ J. Sports Sci., vol. 27, no. 12, pp. 1311–1321, Oct. 2009

    2009

  49. [57]

    McAfee, Table Tennis—Steps to Success

    R. McAfee, Table Tennis—Steps to Success. Human Kinetics, Inc., 2009, pp. 14–15

    2009

  50. [58]

    A. M. Williams and P . R. Ford, ‘‘Expertise and expert performance in sport,’’ Int. Rev. Sport Exercise Psychol., vol. 1, no. 1, pp. 4–18, Feb. 2008

    2008

  51. [59]

    Abernethy and D

    B. Abernethy and D. G. Russell, ‘‘Expert-novice differences in an applied selective attention task,’’ J. Sport Psychol., vol. 9, no. 4, pp. 326–345, Dec. 1987

    1987

  52. [60]

    Renshaw, K

    I. Renshaw, K. Davids, and G. J. P . Savelsbergh, Motor Learning in Practice: A Constraints-Led Approach. Evanston, IL, USA: Routledge, 2012

    2012

  53. [61]

    D. W.-C. Wong, W. C.-C. Lee, and W.-K. Lam, ‘‘Biomechanics of table tennis: A systematic scoping review of playing levels and maneuvers,’’ Appl. Sci., vol. 10, no. 15, p. 5203, Jul. 2020

    2020

  54. [62]

    Han, ‘‘The influence of speed quality training on the development of table tennis,’’ in Proc

    L. Han, ‘‘The influence of speed quality training on the development of table tennis,’’ in Proc. Int. Sports Sci. Educ. Technol. Conf. (ISSETC), Dec. 2019, pp. 1–15

    2019

  55. [63]

    I. M. Lanzoni, S. Bartolomei, R. Di Michele, Y . Gu, J. S. Baker, S. Fantozzi, and M. Cortesi, ‘‘Kinematic analysis of the racket position during the table tennis top spin forehand stroke,’’ Appl. Sci., vol. 11, no. 11, p. 5178, Jun. 2021

    2021

  56. [64]

    Meyer, ‘‘Transformation geometry III: Similarity, inversion, and projection,’’ in Geometry and its Applications, 2nd ed., Burlington, NJ, USA: Academic, 2006, ch

    W. Meyer, ‘‘Transformation geometry III: Similarity, inversion, and projection,’’ in Geometry and its Applications, 2nd ed., Burlington, NJ, USA: Academic, 2006, ch. 7, pp. 311–377

    2006

  57. [65]

    van der Maaten and G

    L. van der Maaten and G. E. Hinton, ‘‘Visualizing data using t-SNE,’’ J. Mach. Learn. Res., vol. 9, no. 86, pp. 2579–2605, Jan. 2008. 25444 VOLUME 13, 2025 M. Pandukabhaya et al.: Performance Benchmarking of Psychomotor Skills Using Wearable Devices MAHELA PANDUKABHAYA (Member...

    2008

Pith tools

Reviewed August 12, 2026 · model on record in the stance chip above.