Pith. sign in

REVIEW 4 major objections 6 minor 42 references

Embodied Tactile Perception of Soft Objects Properties

T0 review · 4 major / 6 minor · reviewed 2026-08-05 · deepseek-v4-flash

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

Pith's one-line read This paper argues that a modular e-Skin that encodes force and vibration in separate pathways and fuses them late, within an action-conditioned latent state-space model, estimates soft object properties more accurately than any single sensi

desk verdict A genuinely useful systematic ablation of how skin stiffness, sensing modality, and palpation strategy shape tactile soft-object estimation, wrapped in a latent-filter model whose generalizable/causal claims outrun the sequence-level evaluation protocol. read the letter →

arxiv 2508.09836 v1 pith:YEODGOQS submitted 2025-08-13 cs.RO

Anirvan Dutta , Alexis WM Devillard , Zhihuan Zhang , Xiaoxiao Cheng , Etienne Burdet This is my paper · ORCID
classification cs.RO
keywords tactileperceptione-Skinmultimodalsensingdeepstate-spacemodellatentfilterpalpationsoftroboticsviscoelasticity
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

This paper sets out to show that tactile perception of soft objects is a joint product of the sensing skin's mechanics, the sensor channels it carries, and the way the robot moves, and that all three can be unified in one learning model. The authors build a modular e-Skin whose silicone stiffness can be switched between soft and hard and whose readings include normal forces, shear-proxy forces, and vibrations, then palpate a set of wave objects with controlled stiffness, surface texture, and heterogeneity using pressing, precession, and sliding. They propose the Latent Filter, an unsupervised, action-conditioned deep state-space model that folds temporal interaction dynamics into a structured latent space, and show that a late-fusion combination of force and vibration channels yields the lowest normalized mean squared error across all interaction primitives. The result matters because robots that manipulate deformable objects—in medical palpation, food handling, or delicate assembly—need to infer physical properties from active touch rather than from pre-labelled categories.

What carries the argument

The Latent Filter is an unsupervised, action-conditioned deep state-space model that factorizes the latent state into a directly observable component $z_t$ and an indirectly observable component $y_t$, with an LSTM approximating the latter, Bayesian integration of an inverse measurement model with a recursive dynamics model, and a learnable hierarchical prior conditioned on action and object label that pushes the latents toward causal mechanical properties. Its companion is the modular e-Skin: an accelerometer array for high-frequency vibration and two FSR arrays separated by a compliant interlayer for normal and differential (shear-proxy) forces, in soft (Ecoflex) or hard (DragonSkin) silic

What would settle it

Retrain the latent filter and the kernel ridge regressor on a random subset of objects (e.g., all sequences from objects 0-13) and test on the remaining objects (14-31, including the heterogeneous set), reporting NMSE for MULTI-L against single modalities. If the late-fusion advantage disappears or errors rise sharply compared with the reported sequence-split numbers, the generalization and modality-superiority claims are artefacts of object leakage. A second decisive check is to shuffle object labels while keeping the same tactile sequences: if the latent space still separates objects rather

Watch

Extended reading notes

Core claim

The paper's central result is that the late-fusion multimodal configuration (MULTI-L), in which the accelerometer layer and the two force-sensing-resistor layers are encoded by separate networks and combined only at the latent level, consistently achieves the lowest normalized mean squared error for estimating the wave objects' spatial frequency, amplitude, stiffness, and heterogeneity, across pressing, precession, and sliding. Two further findings carry the argument: the two FSR layers are the main contributors for stiffness and heterogeneity because differential normal forces proxy shear and skin stretch, while the accelerometer contributes transient information that helps most in sliding;

Load-bearing premise

The load-bearing premise is that evaluating the model on held-out interaction sequences of the same physical objects measures generalization; if instead the objects themselves must be held out, the model may be memorizing object-specific latent patterns rather than learning the mechanical properties it claims to infer.

Editorial extensions

If this is right

  • Late fusion of force and vibration channels, rather than early concatenation, should be the default architecture for multimodal tactile encoders in soft-robot perception.
  • Palpation trajectories should be chosen by target property: moderate frequencies around 0.6 Hz with sufficient indentation depth improve estimation, while pressing is relatively insensitive to parameter choice beyond convergence speed.
  • Skin stiffness should be co-designed with task: softer skins for heterogeneous or compliant objects and sliding interactions, harder skins for bulk stiffness and surface geometry.
  • Unsupervised, action-conditioned latent dynamics can replace handcrafted features and fixed category classifiers for soft-object property regression, enabling continuous estimation of multiple physical properties at once.

Reading between the lines

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

  • A directly testable extension is an object-split evaluation: train the latent filter on a subset of the objects and test on unseen objects. Because the current 75/25 split separates interaction sequences rather than objects, such an evaluation would show whether the late-fusion advantage and the causal-claim survive truly novel objects.
  • The causal reading of the latent dimensions depends on the supervised alignment used for evaluation; a natural extension is to probe latent traversals directly and check whether they monotonically track each generative parameter without object identity.
  • The specific advantage of soft skin on heterogeneous objects suggests an adaptive-compliance skin that stiffens or softens in real time could outperform either fixed configuration across the full object set.
  • A multi-skin comparison with varied silicone thicknesses or artificial fingerprints could separate the contribution of mechanical filtering from the sensor channel itself, which the current two-point stiffness comparison cannot resolve.
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 / 6 minor

Summary. The paper studies how sensor embodiment, sensing modality, and interaction strategy jointly shape tactile perception in robots. It introduces a modular e-Skin with two stiffness variants (soft Ecoflex, hard DragonSkin) and three sensing channels (accelerometers, two FSR layers), a curated set of 32 wave objects with controlled amplitude, spatial frequency, stiffness, and heterogeneity, and three palpation primitives (pressing, precession, sliding). The central modeling contribution is the "Latent Filter," an action-conditioned deep state-space model whose latent state is partitioned into directly observable (z_t) and indirectly observable (y_t) parts, trained with a variational ELBO. The learned latent representations are evaluated with kernel ridge regression to predict ground-truth object properties, reporting NMSE over time for different modality configurations, interaction parameters, and skin stiffnesses. The main claims are that late-fusion multimodal encoding (MULTI-L) consistently yields the lowest NMSE, that moderately high interaction frequencies and larger indentation depths improve inference, and that soft versus hard skin offers complementary advantages depending on object softness and interaction type.

Significance. If the empirical claims survive a more rigorous evaluation, the paper would make a useful contribution: it provides a systematic, hardware-grounded study of how skin compliance, multimodal sensing, and action choice affect tactile perception, and it proposes a latent-variable modeling framework that goes beyond static feature extraction. The curated object set and the release of code and a representative data subset are strengths. The time-resolved NMSE analysis is a constructive way to compare sensing strategies. However, the current evaluation protocol does not establish the central generalization and causality claims: the train/test split is not object-disjoint, the hierarchical prior uses object labels during latent-filter training, and the regression evaluation appears to train and test on the same interaction sequences. The claimed superiority of MULTI-L and the interaction-parameter conclusions rest on this protocol, so they are not yet convincing.

major comments (4)
  1. [Section 2.6 / Section 4.2] The evaluation protocol is the main weakness. Section 4.2 states only that "Twenty-five percent of the dataset was reserved for testing," but the split is at the level of interaction sequences, not objects. With only 32 objects and 16 action settings per object, the same physical objects appear in both training and test folds. The kernel-ridge regressor in Section 2.6 can therefore latch onto object-specific latent patterns (effectively object identity) rather than the intended generative factors (spatial frequency, amplitude, stiffness). This undermines the abstract's claims of "generalizable and causal inference" and the Section 2.6 conclusion that MULTI-L is the best sensing strategy. Moreover, the regression methodology described in Section 2.6 -- training on the final segments of each interaction sequence and applying to samples from the evolving latent space -- suggests that test s
  2. [Section 2.3 / Abstract] The model is called "unsupervised" in the abstract, introduction, and Section 2.3, but the learnable hierarchical prior p(y_t | a_t, N) explicitly conditions on N, described as "the soft-object label." This means object identity is available during latent-filter training. The model may therefore be label-conditioned or semi-supervised rather than unsupervised. This is not merely a terminological issue: the label-conditioned prior could allow the latent variable y_t to encode object identity directly, which is directly relevant to the leakage concern above. The claim should be revised, and an ablation without the N-conditioned prior should be reported to support the "unsupervised" and "causal" interpretation.
  3. [Table 2 / Section 2.8] The quantitative comparisons lack statistical support. Figure 5 shows NMSE curves with standard deviation bands, but the caption states that the standard deviation is scaled by a factor of 10 "for visual clarity" without justification, and no number of seeds or repeated runs is reported. Tables 1 and 2 report single NMSE values per action parameter and per object group with no confidence intervals or significance tests. Consequently, the claimed consistent advantage of MULTI-L over FSRtb, and the differences between interaction parameters, could be within run-to-run variability. I request repeated training runs (or at least bootstrap confidence intervals over sequences/objects) and significance testing for the central comparisons.
  4. [Table 2 / Section 2.8] Table 2 shows that for heterogeneous objects, the NMSE for heterogeneity is very high (e.g., 0.516, 0.376, 0.629 for soft skin; 0.543, 0.685, 1.172 for hard skin), yet the text emphasizes that soft skin is "particularly well-suited for perceiving the properties of heterogeneous materials." An NMSE above 0.5 for the very property that defines the heterogeneous class suggests the model fails to estimate heterogeneity reliably. This tension should be discussed explicitly; the current narrative overstates the positive result and underplays a clear failure case.
minor comments (6)
  1. [Eq. (3) and Eq. (13)] The ELBO expressions contain unmatched brackets and the KL terms are written without closing parentheses: e.g., "KL[qf ilt(zt)||p(zt|zt−1, yt, at)" is missing a closing parenthesis. This makes the derivation harder to follow.
  2. [Section 2.5] Typo: "ACC can detect features by sliding that FSRs alone cannot not distinguish" should be "cannot distinguish."
  3. [Section 3] The sentence "The tactile perception through sensors embodiment and the latent filter" is incomplete and should be rewritten.
  4. [Figure 5 caption] The scaling of standard deviations by a factor of 10 is not explained. If the bands are small, report them unscaled or use a log scale; if they are scaled for visibility, state why and indicate the true magnitude.
  5. [Section 4.2] The preprocessing timing is inconsistent: accelerometer data is said to be downsampled to 600 Hz yielding 6000 samples, while FSR data is downsampled to 30 Hz yielding 300 samples. The ratio is 20x, which is consistent, but the text should clarify that the 6000 and 300 refer to the same 10-second window. Also, "hop length of 20" in the spectrogram is unusual given a window size of 800; please define units.
  6. [General] Some references are incompletely formatted (e.g., missing publisher location in [4]) and the GitHub repository URL is not checked. Please ensure all references are complete and the repository is accessible.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity: the property predictions are genuine regressions from learned latent features, not identities or fitted labels.

full rationale

The paper's central claim (MULTI-L yields lowest NMSE) is an empirical comparison of downstream kernel-ridge regressions on latent representations produced by the latent filter. The latent filter is trained with a variational ELBO on tactile observations and actions; object properties are not part of its reconstruction objective. The downstream regressor is trained on ground-truth properties and then evaluated on latent samples, so the reported NMSE is a real measure of mapability, not a tautology. Although the hierarchical prior p(yt|at,N) conditions on the soft-object label N (Section 2.3), which weakens the 'unsupervised' wording and may allow object identity to be encoded, this does not make the property predictions equal to inputs by construction; a separate regressor must still learn the continuous property mapping. The main self-citations ([8] e-Skin hardware, [31] prior method) are not load-bearing theoretical premises. The main validity caveat is the data-split protocol (Section 4.2: 25% reserved for testing, at sequence rather than object level), which may inflate generalization claims, but this is an experimental-design issue, not circularity. Therefore no circular step can be identified from the paper's equations or citations.

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

The central claims rest on several modeling choices: the Markov assumption, the z/y observability partition, the decodability of properties from the latent space, and the default hardware assumptions of the e-Skin. No physically new entities are postulated. The free parameters are mostly standard hyperparameters, but latent dimension and preprocessing choices are hand-selected and could influence the comparative results.

free parameters (4)
  • latent_dimension_ns = 32
    Number of latent dimensions (16 for z, 16 for y), chosen by hand via hyperparameter tuning, not derived from first principles; affects model capacity.
  • learning_rate = 1e-5
    Adam learning rate selected by tuning; standard hyperparameter.
  • ELBO_annealing_temperature = not specified
    Temperature parameter for annealing the KL weight in the ELBO, tuned manually; affects disentanglement and training stability.
  • spectrogram_window_hop = 800/20
    Preprocessing choices for log-mel spectrogram (window size 800, hop 20, 128 FFT, 49 mel bins); arbitrary but standard choices that could affect results.
assumptions (4)
  • domain assumption First-order Markov assumption on latent states
    Introduced in Supplementary Sec 1 (Eq 6-7) to make inference tractable; loses long-range temporal dependencies that could be relevant to tactile interactions.
  • ad hoc to paper Observability partition: z is directly observable from a single observation, y requires multiple time steps
    Structural assumption in Sec 2.3, motivated by control theory but not validated against alternative partitions; underpins the entire ELBO derivation and the model's ability to separate directly and indirectly observable factors.
  • domain assumption Object properties are decodable from the latent space via kernel ridge regression
    The evaluation in Sec 2.6 assumes that mechanical properties are, in a kernel space, linear functions of the latent variables; if this does not hold, NMSE numbers are not a valid measure of perceptual fidelity.
  • domain assumption Sensor readings faithfully capture the causal physical properties of the object
    The entire study presupposes that e-Skin force and vibration measurements contain the information needed to infer stiffness, texture, and heterogeneity. This is the standard premise of tactile sensing.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Embodied Tactile Perception of Soft Objects Properties." pith.science (2026). https://pith.science/paper/YEODGOQS

@misc{pith2026250809836,
  author       = {Pith},
  title        = {Pith review of: Embodied Tactile Perception of Soft Objects Properties},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/YEODGOQS}},
  note         = {Machine review of arXiv:2508.09836}
}
read the original abstract

To enable robots to develop human-like fine manipulation, it is essential to understand how mechanical compliance, multi-modal sensing, and purposeful interaction jointly shape tactile perception. In this study, we use a dedicated modular e-Skin with tunable mechanical compliance and multi-modal sensing (normal, shear forces and vibrations) to systematically investigate how sensing embodiment and interaction strategies influence robotic perception of objects. Leveraging a curated set of soft wave objects with controlled viscoelastic and surface properties, we explore a rich set of palpation primitives-pressing, precession, sliding that vary indentation depth, frequency, and directionality. In addition, we propose the latent filter, an unsupervised, action-conditioned deep state-space model of the sophisticated interaction dynamics and infer causal mechanical properties into a structured latent space. This provides generalizable and in-depth interpretable representation of how embodiment and interaction determine and influence perception. Our investigation demonstrates that multi-modal sensing outperforms uni-modal sensing. It highlights a nuanced interaction between the environment and mechanical properties of e-Skin, which should be examined alongside the interaction by incorporating temporal dynamics.

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

42 extracted references · 40 canonical work pages

  1. [1]

    Soft Matter 8(25), 6675–6679 (2012)

    Rawlings, A.E., Bramble, J.P., Staniland, S.S.: Innovation through imitation: biomimetic, bioinspired and biokleptic research. Soft Matter 8(25), 6675–6679 (2012)

    work page 2012

  2. [2]

    Hayward, V.: Is there a ‘plenhaptic’ function? Philosophical Transactions of the Royal Society B: Biological Sciences 366(1581), 3115–3122 (2011)

    work page 2011

  3. [3]

    Scientific reports 8(1), 196 (2018)

    Zopf, R., Polito, V., Moore, J.: Revisiting the link between body and agency: visual movement congruency enhances intentional binding but is not body- specific. Scientific reports 8(1), 196 (2018)

    work page 2018

  4. [4]

    In: Human Haptic Perception: Basics and Applications, pp

    Hayward, V.: Haptic shape cues, invariants, priors and interface design. In: Human Haptic Perception: Basics and Applications, pp. 381–392. Springer, ??? (2008)

    work page 2008

  5. [5]

    Frontiers in Robotics and AI 9, 930405 (2022)

    Costi, L., Maiolino, P., Iida, F.: How the environment shapes tactile sens- ing: Understanding the relationship between tactile filters and surrounding environment. Frontiers in Robotics and AI 9, 930405 (2022)

    work page 2022

  6. [6]

    IEEE Robotics and Automation Letters 7(3), 5928–5933 (2022)

    Costi, L., Tagliabue, A., Maiolino, P., Clemens, F., Iida, F.: Magneto-active elas- tomer filter for tactile sensing augmentation through online adaptive stiffening. IEEE Robotics and Automation Letters 7(3), 5928–5933 (2022)

    work page 2022

  7. [7]

    Frontiers in Robotics and AI 8, 665030 (2021)

    Hughes, J., Scimeca, L., Maiolino, P., Iida, F.: Online morphological adaptation for tactile sensing augmentation. Frontiers in Robotics and AI 8, 665030 (2021)

    work page 2021

  8. [8]

    In: IEEE World Haptics Conference (WHC) (2025)

    Devillard, A., Dutta, A., Zhang, Z., Cheng, X., Kaboli, M., Burdet, E.: Modular, multi-layer e-skin for robotics investigations and applications. In: IEEE World Haptics Conference (WHC) (2025). Accepted for publication

    work page 2025

Show all 42 references
  1. [9]

    IEEE Robotics and Automation Letters 7(4), 11314–11321 (2022) 17

    Qiu, P.Z., Tan, Y., Thompson, O., Cobley, B., Nanayakkara, T.: Soft tissue char- acterisation using a novel robotic medical percussion device with acoustic analysis and neural networks. IEEE Robotics and Automation Letters 7(4), 11314–11321 (2022) 17

    2022

  2. [10]

    Soft Robotics 9(2), 280–292 (2022) https://doi.org/10.1089/soro.2020.0129

    Scimeca, L., Hughes, J., Maiolino, P., He, L., Nanayakkara, T., Iida, F.: Action Augmentation of Tactile Perception for Soft-Body Palpation. Soft Robotics 9(2), 280–292 (2022) https://doi.org/10.1089/soro.2020.0129 . Accessed 2024-07-04

    2022

  3. [11]

    IEEE Robotics and Automation Letters 3(2), 1168–1175 (2018) https://doi.org/10.1109/LRA.2018.2793961

    Herzig, N., Maiolino, P., Iida, F., Nanayakkara, T.: A Variable Stiffness Robotic Probe for Soft Tissue Palpation. IEEE Robotics and Automation Letters 3(2), 1168–1175 (2018) https://doi.org/10.1109/LRA.2018.2793961 . Accessed 2024- 07-04

    2018

  4. [12]

    Biomimetics 8(1), 86 (2023) https: //doi.org/10.3390/biomimetics8010086

    Sun, T., Zhang, Z., Miao, Z., Zhang, W.: A Recognition Method for Soft Objects Based on the Fusion of Vision and Haptics. Biomimetics 8(1), 86 (2023) https: //doi.org/10.3390/biomimetics8010086 . Accessed 2024-07-04

    2023 doi

  5. [13]

    IEEE Transactions on Haptics 14(3), 603–614 (2021)

    Cavdan, M., Doerschner, K., Drewing, K.: Task and material properties interac- tively affect softness explorations along different dimensions. IEEE Transactions on Haptics 14(3), 603–614 (2021)

    2021

  6. [14]

    In: Proceedings of the 2023 CHI Conference on Human Factors in Computing Systems, pp

    Steer, C., Dinca, T., Jicol, C., Proulx, M.J., Alexander, J.: Feel the force, see the force: Exploring visual-tactile associations of deformable surfaces with colours and shapes. In: Proceedings of the 2023 CHI Conference on Human Factors in Computing Systems, pp. 1–13 (2023)

    2023

  7. [15]

    In: Di Luca, M

    Visell, Y., Okamoto, S.: Vibrotactile Sensation and Softness Perception. In: Di Luca, M. (ed.) Multisensory Softness: Perceived Compliance from Multiple Sources of Information, pp. 31–47. Springer, London (2014). https://doi.org/10. 1007/978-1-4471-6533-0 3

    2014

  8. [16]

    Future Generation Computer Systems 148, 211–224 (2023) https://doi.org/10.1016/j.future.2023.06.002

    Amin, Y., Gianoglio, C., Valle, M.: Embedded real-time objects’ hardness classi- fication for robotic grippers. Future Generation Computer Systems 148, 211–224 (2023) https://doi.org/10.1016/j.future.2023.06.002 . Accessed 2024-07-04

    2023 doi

  9. [17]

    Vision Research 50(24), 2775–2782 (2010) https://doi.org/10.1016/j.visres.2010.10.005

    Bergmann Tiest, W.M.: Tactual perception of material properties. Vision Research 50(24), 2775–2782 (2010) https://doi.org/10.1016/j.visres.2010.10.005 . Accessed 2025-05-21

    2010 doi

  10. [18]

    Frontiers in integrative neuroscience 12, 59 (2018)

    Lezkan, A., Metzger, A., Drewing, K.: Active haptic exploration of softness: Indentation force is systematically related to prediction, sensation and motiva- tion. Frontiers in integrative neuroscience 12, 59 (2018)

    2018

  11. [19]

    In: Proc

    Xu, Z., Wu, J., Zeng, A., Tenenbaum, J., Song, S.: Densephysnet: Learning dense physical object representations via multi-step dynamic interactions. In: Proc. Robotic. Science and Systems (2019)

    2019

  12. [20]

    Autonomous Robots 44(8), 1377–1393 (2020) https://doi.org/10.1007/s10514-020-09931-y

    Scimeca, L., Maiolino, P., Bray, E., Iida, F.: Structuring of tactile sensory information for category formation in robotics palpation. Autonomous Robots 44(8), 1377–1393 (2020) https://doi.org/10.1007/s10514-020-09931-y . Accessed 2024-07-04 18

    2020 doi

  13. [21]

    In: International Conference on Human Haptic Sensing and Touch Enabled Computer Applications, pp

    Hui, J.C., Kuchenbecker, K.J.: Evaluating the biotac’s ability to detect and char- acterize lumps in simulated tissue. In: International Conference on Human Haptic Sensing and Touch Enabled Computer Applications, pp. 295–302 (2014). Springer

    2014

  14. [22]

    PLOS ONE 12(2), 0171706 (2017) https://doi.org/10.1371/journal.pone.0171706

    Konstantinova, J., Cotugno, G., Dasgupta, P., Althoefer, K., Nanayakkara, T.: Palpation force modulation strategies to identify hard regions in soft tissue organs. PLOS ONE 12(2), 0171706 (2017) https://doi.org/10.1371/journal.pone.0171706 . Accessed 2024-07-04

    2017 doi

  15. [23]

    IEEE Robotics and Automation Letters 7(1), 96–103 (2022) https://doi.org/10.1109/LRA.2021.3119393

    Wang, L., Li, Q., Lam, J., Wang, Z.: Tactual Recognition of Soft Objects From Deformation Cues. IEEE Robotics and Automation Letters 7(1), 96–103 (2022) https://doi.org/10.1109/LRA.2021.3119393 . Accessed 2024-05-16

    2022

  16. [24]

    In: 2019 WRC Sympo- sium on Advanced Robotics and Automation (WRC SARA), pp

    Xu, S., Lin, N., Fan, R., Wu, P., Chen, X.: Exploring Hardness and Geometry Information through Active Perception. In: 2019 WRC Sympo- sium on Advanced Robotics and Automation (WRC SARA), pp. 32–37. IEEE, Beijing, China (2019). https://doi.org/10.1109/WRC-SARA.2019.8931910 . h...

    2019

  17. [25]

    IEEE Sensors Letters 8(8), 1–4 (2024) https://doi.org/10.1109/LSENS.2024.3431430

    Bassal, H., Abbass, Y., Gianoglio, C., Valle, M.: Hardness Discrimination Using Piezoelectric-Based Biomimetic Tactile Sensor and Machine Learning. IEEE Sensors Letters 8(8), 1–4 (2024) https://doi.org/10.1109/LSENS.2024.3431430 . Accessed 2025-03-03

    2024

  18. [26]

    IEEE Transactions on Haptics 12(4), 438–450 (2019) https: //doi.org/10.1109/TOH.2019.2924212

    Caldiran, O., Tan, H.Z., Basdogan, C.: Visuo-Haptic Discrimination of Vis- coelastic Materials. IEEE Transactions on Haptics 12(4), 438–450 (2019) https: //doi.org/10.1109/TOH.2019.2924212 . Accessed 2025-05-21

    2019

  19. [27]

    Soft Robotics 9(6), 1220–1231 (2022) https://doi

    Huang, J., Rosendo, A.: Variable Stiffness Object Recognition with a CNN-Bayes Classifier on a Soft Gripper. Soft Robotics 9(6), 1220–1231 (2022) https://doi. org/10.1089/soro.2021.0105 . Accessed 2024-07-04

    2022

  20. [28]

    Elec- tronics 10(1), 96 (2021) https://doi.org/10.3390/electronics10010096

    Bednarek, M., Kicki, P., Bednarek, J., Walas, K.: Gaining a Sense of Touch Object Stiffness Estimation Using a Soft Gripper and Neural Networks. Elec- tronics 10(1), 96 (2021) https://doi.org/10.3390/electronics10010096 . Accessed 2024-07-04

    2021 doi

  21. [29]

    In: 2023 IEEE International Conference on Soft Robotics (RoboSoft), pp

    Nonaka, T., Abdulali, A., Sirithunge, C., Gilday, K., Iida, F.: Soft robotic tac- tile perception of softer objects based on learning of spatiotemporal pressure patterns. In: 2023 IEEE International Conference on Soft Robotics (RoboSoft), pp. 1–7 (2023). https://doi.org/10.110...

    2023

  22. [30]

    IEEE Transactions on Robotics and Automation 16(5), 496–504 (2000) https://doi.org/10.1109/70.880800

    Bicchi, A., Scilingo, E.P., De Rossi, D.: Haptic discrimination of softness in teleop- eration: the role of the contact area spread rate. IEEE Transactions on Robotics and Automation 16(5), 496–504 (2000) https://doi.org/10.1109/70.880800 . Accessed 2025-05-21 19

    2000 doi

  23. [31]

    IEEE Transactions on Haptics (2023)

    Uttayopas, P., Cheng, X., Eden, J., Burdet, E.: Object recognition using mechanical impact, viscoelasticity, and surface friction during interaction. IEEE Transactions on Haptics (2023)

    2023

  24. [32]

    Journal of the Mechanical Behavior of Biomedical Materials 104, 103630 (2020) https://doi.org/10.1016/j.jmbbm.2020.103630

    Estermann, S.-J., Pahr, D.H., Reisinger, A.: Quantifying tactile properties of liver tissue, silicone elastomers, and a 3D printed polymer for manufacturing realistic organ models. Journal of the Mechanical Behavior of Biomedical Materials 104, 103630 (2020) https://doi.org/10...

    2020

  25. [33]

    In: Ranzato, M., Beygelzimer, A., Dauphin, Y., Liang, P.S., Vaughan, J.W

    Klushyn, A., Kurle, R., Soelch, M., Cseke, B., Smagt, P.: Latent matters: Learning deep state-space models. In: Ranzato, M., Beygelzimer, A., Dauphin, Y., Liang, P.S., Vaughan, J.W. (eds.) Advances in Neural Information Processing Systems, vol. 34, pp. 10234–10245. Curran Asso...

    2021

  26. [34]

    Journal of Open Source Software 3(29) (2018)

    McInnes, L., Healy, J., Saul, N., Großberger, L.: Umap: Uniform manifold approximation and projection. Journal of Open Source Software 3(29) (2018)

    2018

  27. [35]

    In: Proceedings of International Conference on Learning Representations (ICLR) (2017)

    Karl, M., Soelch, M., Bayer, J., Smagt, P.: Deep variational bayes filters: Unsu- pervised learning of state space models from raw data. In: Proceedings of International Conference on Learning Representations (ICLR) (2017)

    2017

  28. [36]

    In: Machine Learning: A Probabilistic Perspective, pp

    Murphy, K.P.: Kernel ridge regression. In: Machine Learning: A Probabilistic Perspective, pp. 492–493. The MIT Press, Cambridge, MA (2012). Chap. 14.4.3

    2012

  29. [37]

    Science 298(5598), 1627–1630 (2002)

    Hillis, J.M., Ernst, M.O., Banks, M.S., Landy, M.S.: Combining sensory infor- mation: mandatory fusion within, but not between, senses. Science 298(5598), 1627–1630 (2002)

    2002

  30. [38]

    Acta psychologica 84(1), 29–40 (1993)

    Lederman, S.J., Klatzky, R.L.: Extracting object properties through haptic exploration. Acta psychologica 84(1), 29–40 (1993)

    1993

  31. [39]

    In: 2019 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), pp

    Moulard, T., Woodall, W., Thomas, D., et al.: Ros 2: Towards a next generation robotics middleware. In: 2019 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), pp. 8703–8709 (2019). https://doi.org/10.1109/ IROS40897.2019.8967700

    2019

  32. [40]

    https://developers.google.com/ protocol-buffers

    Google Open Source: Protocol Buffers. https://developers.google.com/ protocol-buffers. Accessed: 2025-07-31 (2025)

    2025

  33. [41]

    In: SciPy, pp

    McFee, B., Raffel, C., Liang, D., Ellis, D.P., McVicar, M., Battenberg, E., Nieto, O.: librosa: Audio and music signal analysis in python. In: SciPy, pp. 18–24 (2015)

    2015

  34. [42]

    In: Advances in Neural Information Processing Systems (NeurIPS), pp

    Paszke, A., Gross, S., Massa, F., Lerer, A., Bradbury, J., Chanan, G., Killeen, T., Lin, Z., Gimelshein, N., Antiga, L., et al.: Pytorch: An imperative style, high- performance deep learning library. In: Advances in Neural Information Processing Systems (NeurIPS), pp. 8024–803...

    2019

Pith tools

Reviewed August 5, 2026 · model on record in the stance chip above.