Pith. sign in

REVIEW 4 major objections 5 minor 53 references

Estimating Perceptual Attributes of Haptic Textures Using Visuo-Tactile Data

T0 review · 4 major / 5 minor · reviewed 2026-08-07 · deepseek-v4-flash

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

Pith's one-line read A two-stream network fusing camera images and tool vibrations predicts human ratings of surface textures—rough/smooth, flat/bumpy, sticky/slippery, hard/soft—with mean errors of 4.5–6.7 on a 100-point scale, beating single-modality…

desk verdict A plausible incremental visuo-tactile haptic attribute predictor whose qualitative conclusion is probably right, but whose reported error numbers are not yet trustworthy. read the letter →

arxiv 2505.16352 v1 pith:O5Y4VVWS submitted 2025-05-22 cs.HC

Mudassir Ibrahim Awan , Seokhee Jeon This is my paper
classification cs.HC
keywords haptictextureattributesvisuo-tactilelearningConvLSTMCNNautoencoderperceptualattributepredictionleave-one-outcross-validationMFCCperception
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 tries to establish that a machine can predict how people perceive surface textures from physical signals, and that combining vision and touch does this better than either modality alone. It builds a dataset of 50 real textures, collects images and tool-recorded vibrations, force, and speed, and obtains human ratings along four bipolar attribute pairs through psychophysical experiments. Under leave-one-out cross-validation, the proposed two-stream network reports mean absolute errors of 5.23, 4.48, 6.67, and 5.21 for rough–smooth, flat–bumpy, sticky–slippery, and hard–soft, with RMSE values of 6.81, 5.67, 7.52, and 6.13, all lower than the tested single-modality baselines. If correct, this offers a scalable substitute for expensive human rating studies and a way to select or compress haptic feedback in VR, teleoperation, and robotic perception.

What carries the argument

The load-bearing object is the visuo-tactile Net, a two-stream architecture. The visual stream (HV-Net) is a CNN-based autoencoder that compresses a 2304-dimensional vector of ResNet-50 and GLCM features into a 128-dimensional representation; the tactile stream (HT-Net) is a six-layer 1D-ConvLSTM, a recurrent network whose internal state transitions use convolutions so it learns spatial and temporal structure together, ingesting per-segment MFCC features from acceleration plus statistical descriptors of speed and force to produce another 128-dimensional representation. The two representations are concatenated into 256 dimensions and passed through fully connected layers to regress the four attribute scores. The ConvLSTM is the mechanism that lets the tactile stream preserve spatial structure while learning temporal dependencies in vibration signals, and the autoencoder is what lets the visual stream denoise and compress high-dimensional image features.

What would settle it

Run the same trained pipeline on ten new textures rated with the same protocol and compare per-attribute MAE and RMSE to the reported values; if the errors climb above the roughly 10-point just-noticeable difference cited in the paper, the claimed generalization to unseen textures would not hold.

Watch

Extended reading notes

Core claim

The paper claims that perceptual ratings of surface texture can be predicted from physical signals by a two-stream deep network, and that combining visual and tactile input is strictly better than using either modality alone. On a dataset of 50 real textures, human ratings along four bipolar attribute pairs (rough–smooth, flat–bumpy, sticky–slippery, hard–soft) serve as targets; images supply the visual stream and tool-recorded acceleration, speed, and force supply the tactile stream. Under leave-one-out cross-validation, the visuo-tactile network reports mean absolute errors of 5.23 (R-S), 4.48 (F-B), 6.67 (S-S), and 5.21 (H-S) and RMSE values of 6.81, 5.67, 7.52, and 6.13, lower on every attribute than the five baselines tested. The authors interpret this as evidence that a perceptually aligned mapping from physical signals to human ratings is learnable and can generalize to textures left out of training.

Load-bearing premise

The headline error numbers rest on the assumption that leave-one-out cross-validation on 50 textures is a trustworthy estimate of performance on new textures, which presumes no information from the held-out texture leaks into preprocessing, feature normalization, or model selection; the paper does not specify how per-segment tactile predictions are aggregated into one texture rating.

Editorial extensions

If this is right

  • A model that predicts four perceptual ratings from physical signals can serve as a fast proxy for psychophysical rating sessions, which are time-consuming and costly.
  • In model-mediated teleoperation, predicted attributes could select or reconstruct haptic feedback from a texture library when raw signals cannot be transmitted in real time.
  • Perception-based compression becomes possible: store or send attribute vectors instead of raw acceleration and image data, then resynthesize feedback from those vectors.
  • The hybrid features (ResNet+GLCM for vision, MFCC for touch) improve over either feature family alone, so feature fusion matters even before network fusion.
  • Reported errors sit mostly below the roughly 10-point just-noticeable difference cited in the paper, meaning predicted ratings would often be perceptually indistinguishable from human ratings.

Reading between the lines

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

  • A direct test the paper does not run: take a fresh set of textures, record only images, and check how much of the multimodal accuracy survives; the authors' own feature table suggests vision alone may already capture flat-bumpy, so the tactile stream's unique contribution may vary by attribute.
  • If the reported errors are reproducible under properly leakage-free leave-one-out evaluation, the same feature pipeline could be turned into a label generator for psychophysical studies, letting researchers rate textures without new human experiments.
  • The framework's dependence on a 50-texture dataset and a lab-grade tool may not transfer to different sensors or bare-finger interaction; testing with consumer-grade hardware would bound its practical reach.
  • The unspecified per-segment aggregation and normalization steps make the reported LOOCV numbers likely optimistic; a user should verify performance on genuinely new textures before relying on the model.
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. This paper proposes a multimodal visuo-tactile deep-learning framework for predicting four perceptual haptic attributes (rough-smooth, flat-bumpy, sticky-slippery, hard-soft) of 50 real-world textures. Visual features are extracted with pre-trained ResNet-50 plus GLCM descriptors and processed by a 1D-CNN autoencoder; tactile features are MFCCs and statistical descriptors of speed and force, processed by a ConvLSTM. The two 128-dimensional streams are fused and regressed to the four attribute ratings. Evaluation uses leave-one-out cross-validation (LOOCV) against several unimodal and multimodal baselines, reporting MAE and RMSE. The authors claim the proposed method outperforms all baselines on every attribute pair, with the lowest MAE values of 5.23 (R-S), 4.48 (F-B), 6.67 (S-S), and 5.21 (H-S). The paper also presents an individual-feature analysis and discusses class-level error patterns.

Significance. If the quantitative claims are supported, the paper would make a useful contribution to haptic texture perception modeling: it applies ConvLSTM to tactile haptic signals for the first time, combines visual and tactile features in a relatively simple fusion architecture, and evaluates on a real psychophysical dataset with 50 physical textures and four rated attribute pairs. The qualitative direction of the result—multimodal fusion outperforming single-modality baselines—is plausible and consistent with the literature. The paper also provides a useful four-dimensional perceptual-space visualization. However, the central quantitative claims currently rest on an incompletely specified evaluation protocol: the LOOCV unit (texture vs. segment vs. patch) is not stated, early stopping appears to require a validation split that is not described, and feature and architecture selection were conducted using all 50 textures. These issues make the reported MAE/RMSE values potentially optimistic, and the absence of uncertainty measures or significance tests means the margins over baselines may not be meaningful.

major comments (4)
  1. [VI-B and VI-C; Eqs. (1)-(2)] The LOOCV evaluation unit is unspecified. Section VI-B says that each iteration trains on 49 textures and validates on one texture, but the input instances are 0.5-second overlapping tactile segments (Section V-A) and augmented image patches (Section V-B). The paper never states whether the errors in Tables 2-3 are computed per segment/patch and pooled, or aggregated to a texture-level prediction, nor what aggregation rule is used (e.g., averaging segment-level predictions). If overlapping segments from the held-out texture appear in the training set, or if per-segment errors are pooled, then Eq. (1) with n=50 in Section VI-A is not a correct texture-level estimator and the reported MAE/RMSE values are optimistically biased. Please specify the exact evaluation unit and aggregation procedure, and ensure no temporal or spatial overlap crosses the train/test boundary.
  2. [III-C with VI-B] Early stopping appears to be incompatible with the described LOOCV protocol. Section III-C states that training is run for up to 200 epochs with early stopping based on validation performance and a patience of 10 epochs. In the LOOCV scheme of Section VI-B, each fold has only one held-out texture; if that texture is used as the validation set for early stopping, test information leaks into model selection. If instead an inner validation split is used, that split is not described anywhere. This issue affects every reported MAE/RMSE value in Tables 2-3. Please clarify which validation data control early stopping, or replace early stopping with a fixed epoch count.
  3. [III-C and VI-E] Feature and architecture selection appear to have been performed using all 50 textures, making the reported results selection-optimized rather than independent. Section III-C says the architecture configuration was determined through 'extensive empirical experiments,' and Section VI-E states that 1D-DWT and DFT were dropped after 'initial trials' because of poor performance. These choices were evidently made under the same LOOCV error metric that is later used to report final performance. With n=50 and no held-out test set or nested cross-validation, the 1-3 point MAE gains over the best tactile baseline in Tables 2-3 may be within selection noise. Please either implement nested LOOCV, or explicitly disclose the selection process and provide selection-robust estimates (e.g., reporting performance for the full feature set as well as the selected set).
  4. [Tables 2-3 and VI-C] No measures of uncertainty or statistical significance are reported. Tables 2 and 3 give single-point MAE/RMSE values with no standard deviations, per-fold errors, random-seed variation, or paired significance tests across textures. For example, the proposed method's 5.23 MAE for R-S is compared with Tactile SVM's 9.40, but the reader cannot tell whether this gap is stable across the 50 LOOCV folds. Please report per-texture error distributions, confidence intervals (e.g., bootstrap), and paired tests (e.g., Wilcoxon signed-rank) for comparisons against the strongest baselines. This is a necessary complement to the LOOCV point estimates.
minor comments (5)
  1. [VI-E and Table 4] The text describing Table 4 appears to misread the table: Section VI-E says the combined visual features achieved an RMSE of 10.11 for R-S and outperformed ResNet (18.29) and GLCM (19.11), but Table 4 shows the concatenated visual features have R-S RMSE 13.26, while 10.11 is the F-B value. Please correct the text or the table.
  2. [IV and I] Section IV explicitly states that the experimental setup and dataset were adopted from the authors' previous study [14], while the abstract and introduction present the psychophysical experiments and data collection as part of this work. Please clarify what data are reused, what are newly collected, and what the novel contribution is relative to [14].
  3. [VII] The claim that class-wise errors fall within the 'Just Noticeable Difference (JND) threshold for perceptual similarity, often estimated at around 10 out of 100 [35]' is not well supported: reference [35] is not a standard source for JND on this type of rating scale, and no experiment here measures JND. Please provide a proper citation or temper the claim.
  4. [Figure 8] Figure 8 plots actual versus predicted attribute values without any indication of inter-participant variability in the averaged ratings; adding per-texture rating spreads or confidence intervals would help assess how much of the residual error is attributable to the model versus human rater noise.
  5. [IV-C] The statement that the HPS is the 'first visualization to consolidate multi-dimensional haptic attributes into a unified 2D framework' is a strong claim; the 'to the best of our knowledge' qualifier is present, but a brief comparison with existing multidimensional visualization approaches in prior haptic studies would strengthen it.

Circularity Check

0 steps flagged · score 1.0 of 10

No construction-level circularity: the multimodal fusion claim rests on a texture-level LOOCV holdout, not on self-citations; remaining concerns are selection-bias risks, not circular reductions.

full rationale

The paper's central claim—that fusing visual (ResNet+GLCM) and tactile (MFCC) features outperforms single-modality baselines—is an empirical claim supported by leave-one-out cross-validation over 50 textures. The evaluation section explicitly defines the folds at the texture level: "For each iteration, 49 textures (n−1) were used for training, while the remaining texture was reserved for validation." Thus the predicted attribute ratings are not defined in terms of the fitted model or vice versa. The perceptual ratings are reused from the authors' prior work ("the experimental setup and dataset were adopted from our previous study [14]"), and the baselines include the authors' own Haptic CNN [14] and CNN-LSTM [13]; these are self-citations, but they are not load-bearing for the fusion result, which is tested on held-out textures. The paper also admits that the architecture was "determined through extensive empirical experiments" and that DWT/DFT features were dropped after "initial trials," and Section VII acknowledges that "LOOCV can introduce biases in certain cases." These are legitimate threats to the unbiasedness of the reported error bars, but they are selection-bias/correctness concerns rather than circular reductions: no equation makes the predicted rating equal to the training label or to a fitted parameter, and no cited uniqueness theorem forces the architecture. The derivation chain from physical features to perceptual ratings is therefore not circular.

Assumptions & free parameters 4 free parameters · 3 assumptions · 0 invented entities

The central claim rests on hand-selected features and architecture choices and on the reliability of averaged human ratings. No new physical entities or theoretical constructs are introduced, so the ledger is dominated by free parameters and domain assumptions.

free parameters (4)
  • architecture hyperparameters (ConvLSTM layer sizes, FC dims, pooling) = not specified
    Chosen through extensive empirical experiments (Sec. III-C), but the search procedure and validation split are not reported, so the choices could be tuned on the same LOOCV folds.
  • MFCC configuration (13 coefficients, 25 ms frames, 50% overlap) = 13 coefficients, 25 ms, 50%
    Hand-selected standard settings; no ablation on these choices is provided.
  • GLCM gray levels and ResNet patch count = 16 gray levels, 49 patches
    Hand-selected; no analysis of sensitivity to these values.
  • adjective relevance threshold = 50%
    Adjectives with relevance scores of 50% or higher were retained (Sec. IV-B), which determines the four attribute pairs used as regression targets.
assumptions (3)
  • domain assumption Participant ratings averaged over 26 people are reliable ground truth for perceptual attributes.
    The regression targets are mean ratings; inter-rater reliability is not reported (Sec. IV-C).
  • domain assumption Physical signals (tool acceleration, force, speed, images) contain sufficient information to predict the four perceptual attributes.
    The entire mapping from Physical Feature Space to Haptic Perceptual Space assumes this (Secs. III and V).
  • domain assumption LOOCV at the texture level gives an unbiased estimate of generalization to unseen textures.
    The paper uses LOOCV but does not describe how per-segment predictions are aggregated or whether normalization is fitted on training folds only (Sec. VI-B).

how reviews work

0 comments
Cite this review

Pith. "Pith review of Estimating Perceptual Attributes of Haptic Textures Using Visuo-Tactile Data." pith.science (2026). https://pith.science/paper/O5Y4VVWS

@misc{pith2026250516352,
  author       = {Pith},
  title        = {Pith review of: Estimating Perceptual Attributes of Haptic Textures Using Visuo-Tactile Data},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/O5Y4VVWS}},
  note         = {Machine review of arXiv:2505.16352}
}
read the original abstract

Accurate prediction of perceptual attributes of haptic textures is essential for advancing VR and AR applications and enhancing robotic interaction with physical surfaces. This paper presents a deep learning-based multi-modal framework, incorporating visual and tactile data, to predict perceptual texture ratings by leveraging multi-feature inputs. To achieve this, a four-dimensional haptic attribute space encompassing rough-smooth, flat-bumpy, sticky-slippery, and hard-soft dimensions is first constructed through psychophysical experiments, where participants evaluate 50 diverse real-world texture samples. A physical signal space is subsequently created by collecting visual and tactile data from these textures. Finally, a deep learning architecture integrating a CNN-based autoencoder for visual feature learning and a ConvLSTM network for tactile data processing is trained to predict user-assigned attribute ratings. This multi-modal, multi-feature approach maps physical signals to perceptual ratings, enabling accurate predictions for unseen textures. To evaluate predictive accuracy, we employed leave-one-out cross-validation to rigorously assess the model's reliability and generalizability against several machine learning and deep learning baselines. Experimental results demonstrate that the framework consistently outperforms single-modality approaches, achieving lower MAE and RMSE, highlighting the efficacy of combining visual and tactile modalities.

Figures

Figures reproduced from arXiv: 2505.16352 by the authors.

Figure 1
Figure 1. FIGURE 1: Overall Framework [PITH_FULL_IMAGE:figures/full_fig_p004_1.png] view at source ↗
Figure 2
Figure 2. FIGURE 2: The proposed visuo-tactile network. It consists of two streams: one for visual data employing an autoencoder and [PITH_FULL_IMAGE:figures/full_fig_p005_2.png] view at source ↗
Figure 3
Figure 3. FIGURE 3: The real texture samples used in this study from diverse categories [PITH_FULL_IMAGE:figures/full_fig_p006_3.png] view at source ↗
Figures from the paper (8 more)
Figure 4
Figure 4. Figure 4: Participants were seated at a table, wearing head [PITH_FULL_IMAGE:figures/full_fig_p007_4.png]
Figure 3
Figure 3. Figure 3: Furthermore, all the texture samples were cut to 100x100 mm for standardization. They were then affixed to pre￾prepared hard acrylic plates of the same dimensions. The acrylic plates, measuring 100x100x5 mm, ensured uniform surface elevation across all samples. Liquid …
Figure 4
Figure 4. Figure 4: FIGURE 4: (a) The perceptual experiment setup. (b) The GUI for adjective ratings experiment. (c) Four-dimensional haptic [PITH_FULL_IMAGE:figures/full_fig_p008_4.png]
Figure 5
Figure 5. Figure 5: It consists of a rigid tool equipped with a detachable [PITH_FULL_IMAGE:figures/full_fig_p008_5.png]
Figure 5
Figure 5. Figure 5: FIGURE 5: Data recording setup. The setup records vibra [PITH_FULL_IMAGE:figures/full_fig_p009_5.png]
Figure 6
Figure 6. Figure 6: FIGURE 6: Acceleration signals for artificial grass recorded [PITH_FULL_IMAGE:figures/full_fig_p009_6.png]
Figure 8
Figure 8. Figure 8: FIGURE 8: Comparison of actual and predicted attributes for 50 textures using the Leave-One-Out Cross-Validation (LOOCV) [PITH_FULL_IMAGE:figures/full_fig_p011_8.png]
Figure 9
Figure 9. Figure 9: FIGURE 9: Heatmap of MAE for four haptic attribute pairs [PITH_FULL_IMAGE:figures/full_fig_p013_9.png]

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

53 extracted references · 50 canonical work pages

  1. [50]

    Application of a convolutional neural network for predicting the occurrence of ventricular tachyarrhyth- mia using heart rate variability features,

    G. T. Taye, H.-J. Hwang, and K. M. Lim, “Application of a convolutional neural network for predicting the occurrence of ventricular tachyarrhyth- mia using heart rate variability features,” Scientific reports, vol. 10, no. 1, p. 6769, 2020. 14 VOLUME 4, 2025 Author et al.: Preparation of Papers for IEEE Access

    work page 2020

  2. [1]

    Large- scale survey on adjectival representation of vibrotactile stimuli,

    Y . Yoo, J. Lee, J. Seo, E. Lee, J. Lee, Y . Bae, D. Jung, and S. Choi, “Large- scale survey on adjectival representation of vibrotactile stimuli,” in Proc. HAPTICS. New York City, United States: IEEE, 2016, pp. 393–395

    work page 2016

  3. [2]

    Learning to feel textures: Predicting perceptual similarities from un- constrained finger-surface interactions,

    B. A. Richardson, Y . Vardar, C. Wallraven, and K. J. Kuchenbecker, “Learning to feel textures: Predicting perceptual similarities from un- constrained finger-surface interactions,” IEEE Transactions on Haptics, vol. 15, no. 4, pp. 705–717, 2022

    work page 2022

  4. [3]

    Consonance of vibrotactile chords,

    Y . Yoo, I. Hwang, and S. Choi, “Consonance of vibrotactile chords,” IEEE transactions on haptics, vol. 7, no. 1, pp. 3–13, 2013

    work page 2013

  5. [4]

    Deep visuo-tactile learning: Estimation of tactile properties from images,

    K. Takahashi and J. Tan, “Deep visuo-tactile learning: Estimation of tactile properties from images,” in 2019 International Conference on Robotics and Automation (ICRA). IEEE, 2019, pp. 8951–8957

    work page 2019

  6. [5]

    Data-driven rendering of anisotropic haptic textures,

    A. Abdulali and S. Jeon, “Data-driven rendering of anisotropic haptic textures,” in Haptic Interaction: Science, Engineering and Design 2. Springer, 2018, pp. 401–407

    work page 2018

  7. [6]

    Modeling and render- ing realistic textures from unconstrained tool-surface interactions,

    H. Culbertson, J. Unwin, and K. J. Kuchenbecker, “Modeling and render- ing realistic textures from unconstrained tool-surface interactions,” IEEE transactions on haptics, vol. 7, no. 3, pp. 381–393, 2014

    work page 2014

  8. [7]

    M. I. Awan, T. Ogay, W. Hassan, D. Ko, S. Kang, and S. Jeon, “Model- mediated teleoperation for remote haptic texture sharing: Initial study VOLUME 4, 2025 13 Author et al.: Preparation of Papers for IEEE Access of online texture modeling and rendering,” in 2023 IEEE International Conference on Robotics and Automation (ICRA). IEEE, 2023

    work page 2025

Show all 53 references
  1. [8]

    Psychophysical dimensions of tactile perception of textures,

    S. Okamoto, H. Nagano, and Y . Yamada, “Psychophysical dimensions of tactile perception of textures,” IEEE Transactions on Haptics, vol. 6, no. 1, pp. 81–93, 2012

    2012

  2. [9]

    Systematic relations between affective and sensory material dimensions in touch,

    K. Drewing, C. Weyel, H. Celebi, and D. Kaya, “Systematic relations between affective and sensory material dimensions in touch,” IEEE Trans- actions on Haptics, vol. 11, no. 4, pp. 611–622, 2018

    2018

  3. [10]

    Visual–tactile fusion object classification method based on adaptive feature weighting,

    P. Zhang, L. Bai, D. Shan, X. Wang, S. Li, W. Zou, and Z. Chen, “Visual–tactile fusion object classification method based on adaptive feature weighting,” International Journal of Advanced Robotic Systems, vol. 20, no. 4, p. 17298806231191947, 2023

    2023

  4. [11]

    Multimodal feature-based surface material classification,

    M. Strese, C. Schuwerk, A. Iepure, and E. Steinbach, “Multimodal feature-based surface material classification,” IEEE transactions on hap- tics, vol. 10, no. 2, pp. 226–239, 2016

    2016

  5. [12]

    Research on texture haptic reconstruction method based on informer model,

    D. Chen, D. Zhu, J. Liu, G. Chen, Y . Fang, and Y . Zhang, “Research on texture haptic reconstruction method based on informer model,” in Proceedings of the 2023 3rd International Conference on Robotics and Control Engineering, 2023, pp. 161–165

    2023

  6. [13]

    Predicting perceptual haptic at- tributes of textured surface from tactile data based on deep cnn-lstm network,

    M. I. Awan, W. Hassan, and S. Jeon, “Predicting perceptual haptic at- tributes of textured surface from tactile data based on deep cnn-lstm network,” in Proceedings of the 29th ACM Symposium on Virtual Reality Software and Technology, 2023, pp. 1–9

    2023

  7. [14]

    Establishing haptic texture attribute space and predicting haptic attributes from image features using 1d-cnn,

    W. Hassan, J. B. Joolee, and S. Jeon, “Establishing haptic texture attribute space and predicting haptic attributes from image features using 1d-cnn,” Scientific Reports, vol. 13, no. 1, p. 11684, 2023

    2023

  8. [15]

    Deep residual learning for image recognition,

    K. He, X. Zhang, S. Ren, and J. Sun, “Deep residual learning for image recognition,” in Proceedings of the IEEE conference on computer vision and pattern recognition, 2016, pp. 770–778

    2016

  9. [16]

    Convolutional lstm network: A machine learning approach for precipi- tation nowcasting,

    X. Shi, Z. Chen, H. Wang, D.-Y . Yeung, W.-K. Wong, and W.-c. Woo, “Convolutional lstm network: A machine learning approach for precipi- tation nowcasting,” Advances in neural information processing systems, vol. 28, 2015

    2015

  10. [17]

    Texture perception through direct and indirect touch: An analysis of perceptual space for tactile textures in two modes of exploration,

    T. Yoshioka, S. J. Bensmaia, J. C. Craig, and S. S. Hsiao, “Texture perception through direct and indirect touch: An analysis of perceptual space for tactile textures in two modes of exploration,” Somatosensory & motor research, vol. 24, no. 1-2, pp. 53–70, 2007

    2007

  11. [18]

    The dimensions of tactile perception of surfaces,

    K. Yoshida et al., “The dimensions of tactile perception of surfaces,” Journal of Texture Studies, vol. 12, pp. 123–135, 1968

    1968

  12. [19]

    Factors contributing to the integration of textural qualities: Evidence from virtual surfaces,

    M. Hollins, F. Lorenz, A. Seeger, and R. Taylor, “Factors contributing to the integration of textural qualities: Evidence from virtual surfaces,” Somatosensory & motor research, vol. 22, no. 3, pp. 193–206, 2005

    2005

  13. [20]

    Perception of the tactile texture of raised-dot patterns: A multidimen- sional analysis,

    G. A. Gescheider, S. J. Bolanowski, T. C. Greenfield, and K. E. Brunette, “Perception of the tactile texture of raised-dot patterns: A multidimen- sional analysis,” Somatosensory & motor research, vol. 22, no. 3, pp. 127– 140, 2005

    2005

  14. [21]

    Softness discrimination with a tool,

    R. H. LaMotte, “Softness discrimination with a tool,” Journal of neuro- physiology, vol. 83, no. 4, pp. 1777–1786, 2000

    2000

  15. [22]

    Ungrounded haptic augmented reality system for displaying roughness and friction,

    H. Culbertson and K. J. Kuchenbecker, “Ungrounded haptic augmented reality system for displaying roughness and friction,” IEEE/ASME Trans- actions on Mechatronics, pp. 1839–1849, 2017

    2017

  16. [23]

    Evaluating differences between bare-handed and tool-based interaction in perceptual space,

    W. Hassan and S. Jeon, “Evaluating differences between bare-handed and tool-based interaction in perceptual space,” in 2016 IEEE Haptics Symposium (HAPTICS). IEEE, 2016, pp. 185–191

    2016

  17. [24]

    Visual and haptic representations of material properties,

    E. Baumgartner, C. B. Wiebel, and K. R. Gegenfurtner, “Visual and haptic representations of material properties,” Multisensory research, vol. 26, no. 5, pp. 429–455, 2013

    2013

  18. [25]

    Robotic learn- ing of haptic adjectives through physical interaction,

    V . Chu, I. McMahon, L. Riano, C. G. McDonald, Q. He, J. M. Perez- Tejada, M. Arrigo, T. Darrell, and K. J. Kuchenbecker, “Robotic learn- ing of haptic adjectives through physical interaction,” Robotics and Au- tonomous Systems, vol. 63, pp. 279–292, 2015

    2015

  19. [26]

    Experimental study on the perception characteristics of haptic texture by multidimensional scaling,

    J. Wu, N. Li, W. Liu, G. Song, and J. Zhang, “Experimental study on the perception characteristics of haptic texture by multidimensional scaling,” IEEE transactions on haptics, pp. 410–420, 2015

    2015

  20. [27]

    Authoring new haptic textures based on interpolation of real textures in affective space,

    W. Hassan, A. Abdulali, and S. Jeon, “Authoring new haptic textures based on interpolation of real textures in affective space,” IEEE transactions on industrial electronics, pp. 667–676, 2019

    2019

  21. [28]

    Surface classification using acceleration signals recorded during human freehand movement,

    M. Strese, C. Schuwerk, and E. Steinbach, “Surface classification using acceleration signals recorded during human freehand movement,” in 2015 IEEE World Haptics Conference (WHC). IEEE, 2015

    2015

  22. [29]

    Preference-driven texture modeling through interactive generation and search,

    S. Lu, M. Zheng, M. C. Fontaine, S. Nikolaidis, and H. Culbertson, “Preference-driven texture modeling through interactive generation and search,” IEEE transactions on haptics, pp. 508–520, 2022

    2022

  23. [30]

    Development and evaluation of a learning-based model for real-time haptic texture rendering,

    N. Heravi, H. Culbertson, A. M. Okamura, and J. Bohg, “Development and evaluation of a learning-based model for real-time haptic texture rendering,” IEEE Transactions on Haptics, 2024

    2024

  24. [31]

    Binding touch to everything: Learn- ing unified multimodal tactile representations,

    F. Yang, C. Feng, Z. Chen, H. Park, D. Wang, Y . Dou, Z. Zeng, X. Chen, R. Gangopadhyay, A. Owens et al., “Binding touch to everything: Learn- ing unified multimodal tactile representations,” in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition,...

    2024

  25. [32]

    Object surface roughness/texture recognition using machine vision enables for human- machine haptic interaction,

    Z. Lin, H. Zheng, Y . Lu, J. Zhang, G. Chai, and G. Zuo, “Object surface roughness/texture recognition using machine vision enables for human- machine haptic interaction,” Frontiers in Computer Science, vol. 6, p. 1401560, 2024

    2024

  26. [33]

    Classification method of visual-tactile fusion dataset based on cnn-tcn,

    H. Li and H. Zhang, “Classification method of visual-tactile fusion dataset based on cnn-tcn,” in 2023 8th International Conference on Control, Robotics and Cybernetics (CRC). IEEE, 2024, pp. 295–300

    2023

  27. [34]

    Texture classification using 2d lstm networks,

    W. Byeon, M. Liwicki, and T. M. Breuel, “Texture classification using 2d lstm networks,” in 2014 22nd international conference on pattern recognition. IEEE, 2014, pp. 1144–1149

    2014

  28. [35]

    Quantifying haptic affection of car door through data-driven analysis of force profile,

    W. Hassan, M. I. Awan, A. Raza, K.-U. Kyung, and S. Jeon, “Quantifying haptic affection of car door through data-driven analysis of force profile,” arXiv preprint arXiv:2411.11382, 2024

    2024 arXiv

  29. [36]

    Attention is all you need,

    A. Vaswani, N. Shazeer, N. Parmar, J. Uszkoreit, L. Jones, A. N. Gomez, Ł. Kaiser, and I. Polosukhin, “Attention is all you need,” Advances in neural information processing systems, vol. 30, 2017

    2017

  30. [37]

    Transformers in time series: A survey,

    Q. Wen, T. Zhou, C. Zhang, W. Chen, Z. Ma, J. Yan, and L. Sun, “Transformers in time series: A survey,” arXiv preprint arXiv:2202.07125, 2022

    2022 arXiv

  31. [38]

    Deep learning approaches for water stress forecasting in arboriculture using time series of remote sensing images: Comparative study between convlstm and cnn- lstm models,

    I. Bounoua, Y . Saidi, R. Yaagoubi, and M. Bouziani, “Deep learning approaches for water stress forecasting in arboriculture using time series of remote sensing images: Comparative study between convlstm and cnn- lstm models,” Technologies, vol. 12, no. 6, p. 77, 2024

    2024

  32. [39]

    A convlstm neural network model for spatiotemporal prediction of mining area surface deformation based on sbas-insar monitoring data,

    S. Yao, Y . He, L. Zhang, W. Yang, Y . Chen, Q. Sun, Z. Zhao, and S. Cao, “A convlstm neural network model for spatiotemporal prediction of mining area surface deformation based on sbas-insar monitoring data,” IEEE Transactions on Geoscience and Remote Sensing, vol. 61, pp. 1–22, 2023

    2023

  33. [40]

    Tensorflow and keras: Open-source deep learning frameworks,

    M. Abadi and F. C. et al., “Tensorflow and keras: Open-source deep learning frameworks,” https://www.tensorflow.org/ and https://keras.io/, 2015, accessed: 2024-03-13

    2015

  34. [41]

    Introduction to multidimensional scaling: Theory, methods, and applications,

    S. Schiffman, “Introduction to multidimensional scaling: Theory, methods, and applications,” 1981

    1981

  35. [42]

    Perceptual space and adjective rating of sinusoidal vibrations perceived via mobile device,

    I. Hwang and S. Choi, “Perceptual space and adjective rating of sinusoidal vibrations perceived via mobile device,” in 2010 IEEE Haptics Sympo- sium. IEEE, 2010, pp. 1–8

    2010

  36. [43]

    Data-driven modeling of anisotropic haptic tex- tures: Data segmentation and interpolation,

    A. Abdulali and S. Jeon, “Data-driven modeling of anisotropic haptic tex- tures: Data segmentation and interpolation,” in International Conference on Human Haptic Sensing and Touch Enabled Computer Applications. Springer, 2016, pp. 228–239

    2016

  37. [44]

    Di- mensional reduction of high-frequency accelerations for haptic rendering,

    N. Landin, J. M. Romano, W. McMahan, and K. J. Kuchenbecker, “Di- mensional reduction of high-frequency accelerations for haptic rendering,” in Haptics: Generating and Perceiving Tangible Sensations: International Conference, EuroHaptics 2010, Amsterdam, July 8-10, 2010. Proce...

    2010

  38. [45]

    H.-G. Kim, N. Moreau, and T. Sikora, MPEG-7 audio and beyond: Audio content indexing and retrieval. John Wiley & Sons, 2006

    2006

  39. [46]

    Textural features for image classification,

    R. M. Haralick, K. Shanmugam, and I. H. Dinstein, “Textural features for image classification,” IEEE Transactions on systems, man, and cybernet- ics, no. 6, pp. 610–621, 1973

    1973

  40. [47]

    Uniform consistency of cross-validation estimators for high-dimensional ridge regression,

    P. Patil, Y . Wei, A. Rinaldo, and R. Tibshirani, “Uniform consistency of cross-validation estimators for high-dimensional ridge regression,” in International Conference on Artificial Intelligence and Statistics. PMLR, 2021, pp. 3178–3186

    2021

  41. [48]

    Comparative analysis of cross-validation techniques: Loocv, k-folds cross-validation, and repeated k-folds cross-validation in machine learning models,

    V . W. Lumumba, D. Kiprotich, N. Makena, M. Kavita, and M. Mpaine, “Comparative analysis of cross-validation techniques: Loocv, k-folds cross-validation, and repeated k-folds cross-validation in machine learning models,” Am. J. Theor. Appl. Stat, vol. 13, pp. 127–137, 2024

    2024

  42. [49]

    Cross-validatory choice and assessment of statistical predic- tions,

    M. Stone, “Cross-validatory choice and assessment of statistical predic- tions,” Journal of the royal statistical society: Series B (Methodological), vol. 36, no. 2, pp. 111–133, 1974

    1974

  43. [51]

    Haptic recognition of texture surfaces using semi-supervised feature learning based on sparse representation,

    Z. Shao, J. Bao, J. Li, and H. Tang, “Haptic recognition of texture surfaces using semi-supervised feature learning based on sparse representation,” Cognitive Computation, pp. 1656–1671, 2023

    2023

  44. [52]

    Wavelet transforms significantly sparsify and compress tactile interactions,

    A. Slepyan, M. Zakariaie, T. Tran, and N. Thakor, “Wavelet transforms significantly sparsify and compress tactile interactions,” Sensors, vol. 24, no. 13, p. 4243, 2024. MUDASSIR IBRAHIM AWAN received his B.E. in Electronics Engineering from the Karachi In- stitute of Economic...

    2024

  45. [2024]

    He is also a co-founder and faculty member in the Department of Metaverse. His research interests include data-driven haptic modeling and rendering, hyper-realistic multimodal feedback in vir- tual, augmented, and remote environments, and the development of modular wearable ha...

    2025

Pith tools

Reviewed August 7, 2026 · model on record in the stance chip above.