Pith. sign in

REVIEW 3 major objections 5 minor 33 references

A 13-DOF three-finger gripper, combining soft tactile fingertips with an actuated vision-tactile palm, achieves dexterous in-hand manipulation previously requiring anthropomorphic hand designs.

Reviewed by Pith at T0; open to challenge. T0 means a machine referee read the full paper against a public rubric. the ladder, T0–T4 →

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

T0 review · deepseek-v4-flash

2026-08-01 23:18 UTC pith:DYHYD7YS

load-bearing objection A useful system paper whose central synergy claim needs ablations the experiments don't provide. the 3 major comments →

arxiv 2607.15448 v1 pith:DYHYD7YS submitted 2026-07-16 cs.RO

VTAP Gripper: Synergizing Fingertip Sensing and a Visuo-Tactile Active Palm for Dexterous In-Hand Manipulation

Yuhao Zhou , Sheeraz Athar , Zhixian Hu , Binghao Huang , Yunzhu Li , Juan Wachs , Yu She This is my paper
classification cs.RO
keywords tactile-reactive grippervisuo-tactile sensingactive palmin-hand manipulationteleoperation retargetingdexterous graspingpeg-in-holeVR hand tracking
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved

The pith

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

The paper aims to show that dexterous in-hand manipulation does not require a high-DOF anthropomorphic robot hand. It builds a 13-DOF three-finger gripper whose active palm slides vertically and doubles as a sensor that can either view the scene or act as a high-resolution tactile skin. Fingertips carry compliant piezoresistive arrays, and a gesture-conditioned retargeting scheme maps human hand motions onto the three-finger structure. Across reactive grasping, syringe reorientation with plunger actuation, singulation of objects as small as 3 mm, and 1 mm-tolerance peg-in-hole insertion, the system reports success rates of 93.3%, 65%, and 70%, respectively. A sympathetic reader would conclude that finger-palm coordination plus multi-modal sensing is a viable design path for the field.

Core claim

The central claim is that a deliberately non-anthropomorphic gripper — three Fin-Ray compliant fingers, each with four actuated degrees of freedom, plus a linearly actuated palm that switches between camera vision and optical tactile sensing via controllable LED illumination — can perform tasks normally used as benchmarks for full hand dexterity. The paper demonstrates a tactile-reactive grasp stop, in-hand syringe reorientation and plunger actuation via teleoperation, singulation of clustered objects down to 3 mm, and autonomous peg-in-hole insertion with 1 mm clearance. The authors attribute this success to structured finger-palm synergy: the palm contributes an extra controllable contact

What carries the argument

The load-bearing mechanism is the actuated bi-modal palm: a linear actuator with a 50 mm stroke moves a palm module containing a camera, a ring of multi-color LEDs, and a silicone elastomer with a reflective coating. With the LEDs off, the module is transparent and acts as a conventional camera for scene localization; with the LEDs on, it becomes an optical tactile sensor by capturing deformation of the reflective surface. This palm adds one degree of freedom and a second contact surface to the gripper's three 4-DOF Fin-Ray fingers, for 13 DOF total. A staged, gesture-conditioned retargeting framework then reduces human-hand mapping to three grasp primitives (cage, power, pinch), constrainin

Load-bearing premise

The demonstrations assume the VR headset's hand tracking is continuously accurate; the paper reports that when the headset moves close enough to capture the little finger, the thumb or index intermittently loses tracking, which the system interprets as a finger release and opens the pinch grasp.

What would settle it

Repeat the syringe teleoperation task with the same retargeting but a data-glove or marker-based hand tracker that has no line-of-sight occlusion; if the success rate stays near 65%, the failures are not primarily caused by VR tracking loss, and the claimed finger-palm dexterity would need to be demonstrated under non-occluded teleoperation.

Watch this falsifier. Get emailed when new claim-graph text bears on it.

Share X Bluesky LinkedIn Reddit HN

If this is right

  • A non-anthropomorphic three-finger gripper can substitute for humanoid hands in tasks like syringe handling and singulation, lowering mechanical complexity and cost.
  • Threshold-based tactile-reactive grasping with fingertip arrays achieved 93.3% success across YCB and fragile objects, suggesting a robust grasping primitive.
  • The bi-modal palm's mode-switching shows that one sensor can serve both pre-contact perception and post-contact tactile feedback without mechanical reconfiguration or stereo vision.
  • The staged retargeting framework offers a practical teleoperation and data-collection interface for heterogeneous three-finger grippers, supporting learning-based manipulation policies.
  • The autonomous vision-tactile peg-in-hole task, with 70% success at 1 mm tolerance, demonstrates precision assembly using palm-based tactile hole localization.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • A natural next experiment the paper does not run is ablating the palm's linear actuation: if disabling it leaves syringe and peg-in-hole success rates unchanged, then finger-palm coordination is not the source of the claimed dexterity.
  • The reported VR-tracking failure mode in the syringe task implies the retargeting framework could gain substantially from a glove-based or marker-based hand tracker; the current success ceiling may be sensor-bound rather than gripper-bound.
  • Because the palm's vision and tactile modes share the same optical path, the observation stream could serve double duty for policy learning: RGB context and dense tactile labels without a separate wrist camera.
  • The singulation task's use of the adduction/abduction DOF suggests that adding a second active axis to the palm (a tilt degree of freedom) would further expand the gripper's in-hand workspace without adding fingers.

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

3 major / 5 minor

Summary. The paper presents the VTAP Gripper, a three-finger, 13-DOF gripper with compliant Fin-Ray fingers, fingertip piezoresistive tactile arrays (FlexiTac), and an actuated palm that switches between camera-based vision and optical tactile sensing via controllable LED illumination. A staged, gesture-conditioned retargeting framework maps Meta Quest 3 VR hand tracking to the non-anthropomorphic gripper for teleoperation. Experiments cover tactile-reactive grasping of nine objects, teleoperated syringe reorientation and plunger actuation, in-hand singulation of objects down to 3 mm, and autonomous vision-tactile peg-in-hole insertion with 1 mm clearance. The paper claims that coordinated finger-palm interaction and multi-modal sensing can achieve high manipulation performance without high-DOF anthropomorphic hands, and that the system is a practical reference architecture for dexterous manipulation and data collection.

Significance. If the central claim were established, this would be a valuable contribution: the hardware integration is thoughtful, the active bi-modal palm is an interesting design, and the retargeting framework addresses a real embodiment gap for non-anthropomorphic grippers. The kinematic derivation is standard and correct, and the experiments cover a diverse set of tasks, including manipulation of very small objects. The paper also gives explicit credit to relevant prior work and openly discusses failure modes. However, the current evidence is largely feasibility-level: all reported successes are whole-system demonstrations, and no control condition isolates the contributions of the active palm or the two sensing modalities. The central 'synergy' claim is therefore not yet supported by the data as presented.

major comments (3)
  1. [§IV (overall), especially §IV-A and §IV-D, and §V] The central claim that high manipulation performance is achieved through coordinated finger-palm interaction and multi-modal sensing is not isolatable from the reported experiments. All task successes (Table I, syringe §IV-B, singulation §IV-C, peg-in-hole §IV-D) are whole-system results. There is no condition with the active palm locked, with the palm's visual or tactile channel disabled, or with fingertip tactile feedback removed. Thus the 93.3% grasping, 65% syringe, and 70% peg-in-hole results could in principle be due to the three Fin-Ray fingers, the teleoperator's skill, or the UR5e arm, with the palm peripheral. Since the paper's novelty and final conclusion (§V) rest on synergy, this is load-bearing. Adding ablations (e.g., palm actuation disabled; vision-only vs. tactile-only palm; tactile threshold open-loop) or explicitly reframing the claim as system-level feasibility would
  2. [§IV-B and §V] The teleoperation evaluation is confounded by the VR input. The paper states in §V that most syringe-task failures were caused by Meta Quest 3 tracking occlusion, which the system interpreted as finger release. Because the retargeting framework is a central contribution, the 13/20 success rate does not isolate the retargeting method's accuracy or the gripper's manipulability; it largely measures VR tracking robustness in this setup. In addition, no baseline retargeting method (e.g., direct joint mapping or the optimization of [26]) and no ablation of the loss terms in Eq. (5) are provided, so the advantage of the staged gesture-conditioned framework is not demonstrated. Please report tracking-failure trials separately, compare against a baseline, or use a less occlusion-prone hand-capture device.
  3. [§IV and Table I] The quantitative evidence for the headline performance numbers is thin. Grasping uses 10 trials per object and the drill is 5/10; syringe uses 20 trials; peg-in-hole uses 10. Only point success rates are reported, without confidence intervals or trial-by-trial variability. Given that the abstract claims 'high manipulation performance' and 'extensive experiments,' the statistical basis is currently modest. Reporting binomial confidence intervals (or increasing trials) and a per-object breakdown for grasping would make the claim commensurate with the evidence.
minor comments (5)
  1. [Eq. (7)] The subscript i is missing on ΔC_f and S_f; clarify that these quantities are computed per finger i before summation.
  2. [§III-C and §IV-A] The retargeting hyperparameters (λ1, λ2, w_j, T_th, α, and the singulation threshold) are not reported. Please provide values and, if possible, a sensitivity analysis, since the method relies on these manually chosen weights.
  3. [Fig. 3(b) and §III-A] The workspace spans 'approximately 469 mm along X, 477 mm along Y, and 311 mm along Z' — clarify whether these are the combined envelope of all three fingertips or the per-finger reach, and specify the origin convention used.
  4. [§IV-C] The comparison 'approximately 215% larger sensing area' than GelSight Mini should cite the exact sensing areas and the GelSight Mini specification for reproducibility.
  5. [§IV-B and Fig. 8] The text says the little-finger contraction commands the palm to depress the plunger, but the retargeting description in §III-C maps only thumb, index, and middle fingers. Explain how little-finger motion is captured and mapped to palm actuation.

Circularity Check

0 steps flagged

No circular reduction; results are empirical demonstrations with standard retargeting optimization, and self-citations are non-load-bearing.

full rationale

The paper's quantitative claims are experimental success rates (Table I, syringe 65%, peg-in-hole 70%) rather than predictions derived from fitted parameters. The retargeting objective (Eqs. 3-5) is a standard weighted optimization with manually chosen weights and a Huber penalty; no quantity is fit to the reported outcomes. The grasping threshold Tth in Eq. 7 is empirically selected, but it is an input to a reactive controller, not a claimed prediction. The central 'finger-palm synergy' claim is supported only by whole-system experiments and lacks ablations (e.g., palm-locked or vision-only conditions); that is an evidence-strength limitation, not a circular step. Self-citations appear (e.g., [17] active-palm gripper, [28] FlexiTac), but the gripper design and experiments stand independently and no argument is forced by those citations. The VR-occlusion failure mode noted in Sec. V is a teleoperation robustness concern, not circularity. No equation or fitted value is equivalent by construction to any reported result.

Axiom & Free-Parameter Ledger

4 free parameters · 4 axioms · 0 invented entities

The central claims rest on standard kinematic modeling plus several domain assumptions about passive compliance, mode-switching optical behavior, and VR tracking reliability. The free parameters are control thresholds and loss weights that are empirically chosen without reported values.

free parameters (4)
  • Tth = not reported
    Termination threshold in Eq. (7); described as 'empirically selected' in Section IV-A, directly controls grasp closure and success rates.
  • alpha = not reported
    Balance between contact-area change and signal magnitude in Eq. (7); no value or sensitivity analysis given.
  • lambda1, lambda2, wj = not reported
    Weights in the retargeting objective Eq. (5); required to reproduce teleoperation, values not provided.
  • Singulation mode distance threshold = not reported
    Predefined fingertip-distance and vector-alignment thresholds that trigger singulation mode (Section III-C); not specified.
axioms (4)
  • domain assumption Fin-Ray TPU passive compliance is sufficient to adapt to object geometry without explicit contact modeling
    Used as design rationale in Section III-A; if false, grasping robustness would depend on unmodeled mechanics.
  • domain assumption The mode-switching palm provides unobstructed external vision when LEDs are off and reliable tactile deformation images when LEDs are on
    Basis for the bi-modal palm claim in Section III-B; no quantitative validation of optical transparency or tactile image quality is provided.
  • domain assumption Meta Quest 3 hand tracking gives keypoints accurate enough for the retargeting optimization
    The entire teleoperation pipeline uses VR input (Section IV-B); the paper reports occlusion causing failures, so this is load-bearing.
  • ad hoc to paper Three grasp primitives (cage, power, pinch) and the thumb-index mapping span the task space used
    Gesture-conditioned subspace reduction (Section III-C) restricts the DOFs to three primitives; there is no proof that this subspace covers other tasks.

pith-pipeline@v1.3.0-alltime-deepseek · 10962 in / 8791 out tokens · 87854 ms · 2026-08-01T23:18:40.755277+00:00 · methodology

0 comments
Cite this review

Pith. "Pith review of VTAP Gripper: Synergizing Fingertip Sensing and a Visuo-Tactile Active Palm for Dexterous In-Hand Manipulation." pith.science (2026). https://pith.science/paper/DYHYD7YS

@misc{pith2026260715448,
  author       = {Pith},
  title        = {Pith review of: VTAP Gripper: Synergizing Fingertip Sensing and a Visuo-Tactile Active Palm for Dexterous In-Hand Manipulation},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/DYHYD7YS}},
  note         = {Machine review of arXiv:2607.15448}
}
Share X Bluesky LinkedIn Reddit HN
read the original abstract

This paper presents a tactile-reactive gripper that integrates a Visuo-Tactile Active Palm (VTAP) and compliant, reconfigurable fingers equipped with tactile array sensors. The design exploits structured finger-palm synergy and multi-modal perception to achieve both robust grasping and fine manipulation. The actuated bi-modal palm seamlessly combines long-range visual localization with contact-rich tactile feedback, substantially extending the system's manipulation capability. To bridge the embodiment gap between human hand motion and the heterogeneous three-finger structure, we further propose a staged, gesture-conditioned retargeting framework for dexterous teleoperation. Extensive experiments validate the system across a range of challenging tasks: reactive grasping of YCB and fragile objects, in-hand syringe reorientation and plunger actuation, singulation of clustered objects down to 3 mm in diameter, and vision-tactile peg-in-hole insertion. Results demonstrate that high manipulation performance can be achieved through coordinated finger-palm interaction and multi-modal sensing, without resorting to high degrees of freedom anthropomorphic designs. The VTAP gripper and its retargeting framework offer a practical reference architecture for dexterous gripper design, manipulation, and contact-rich data collection in support of learning-based approaches. Project webpage: https://yuhochau.github.io/vtap/.

Figures

Figures reproduced from arXiv: 2607.15448 by Binghao Huang, Juan Wachs, Sheeraz Athar, Yuhao Zhou, Yunzhu Li, Yu She, Zhixian Hu.

Figure 1
Figure 1. Figure 1: Overview of the proposed gripper. The system enables in-hand [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: (a) Mechanical design and overall structure of the proposed gripper. [PITH_FULL_IMAGE:figures/full_fig_p002_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: (a) Schematic representation of the finger kinematics. (b) Workspace [PITH_FULL_IMAGE:figures/full_fig_p003_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: (a) Visualization of vision and tactile outputs from the proposed [PITH_FULL_IMAGE:figures/full_fig_p003_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: Overview of the proposed retargeting framework for teleoperated [PITH_FULL_IMAGE:figures/full_fig_p004_5.png] view at source ↗
Figure 6
Figure 6. Figure 6: Tactile-reactive grasping experiments under three representative [PITH_FULL_IMAGE:figures/full_fig_p005_6.png] view at source ↗
Figure 8
Figure 8. Figure 8: In-hand syringe reorientation and plunger actuation via teleoperation [PITH_FULL_IMAGE:figures/full_fig_p006_8.png] view at source ↗
Figure 9
Figure 9. Figure 9: Sequential images of in-hand singulation. (a) Overview of the teleoperated singulation process for grasping granular mung beans from a bowl, [PITH_FULL_IMAGE:figures/full_fig_p007_9.png] view at source ↗
Figure 10
Figure 10. Figure 10: Peg-in-hole insertion enabled by the bi-modal actuated palm. [PITH_FULL_IMAGE:figures/full_fig_p007_10.png] view at source ↗

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

33 extracted references · 2 linked inside Pith

  1. [1]

    Diffusion policy: Visuomotor policy learning via action diffusion,

    C. Chi, Z. Xu, S. Feng, E. Cousineau, Y . Du, B. Burchfiel, R. Tedrake, and S. Song, “Diffusion policy: Visuomotor policy learning via action diffusion,”Int. J. Robot. Res., vol. 44, no. 10-11, pp. 1684–1704, 2025

    2025

  2. [2]

    Design of a fully actuated robotic hand with multiple gelsight tactile sensors,

    A. Wilson, S. Wang, B. Romero, and E. Adelson, “Design of a fully actuated robotic hand with multiple gelsight tactile sensors,”arXiv preprint arXiv:2002.02474, 2020

    Pith/arXiv arXiv 2002

  3. [3]

    The grasp taxonomy of human grasp types,

    T. Feix, J. Romero, H.-B. Schmiedmayer, A. M. Dollar, and D. Kragic, “The grasp taxonomy of human grasp types,”IEEE Trans. Hum. Mach. Syst., vol. 46, no. 1, pp. 66–77, 2016

    2016

  4. [4]

    A hand-centric clas- sification of human and robot dexterous manipulation,

    I. M. Bullock, R. R. Ma, and A. M. Dollar, “A hand-centric clas- sification of human and robot dexterous manipulation,”IEEE Trans. Haptics, vol. 6, no. 2, pp. 129–144, 2012

    2012

  5. [5]

    Actuated palms for soft robotic hands: review and perspectives,

    M. Pozzi, M. Malvezzi, D. Prattichizzo, and G. Salvietti, “Actuated palms for soft robotic hands: review and perspectives,”IEEE/ASME Trans. Mechatron., vol. 29, no. 2, pp. 902–912, 2023

    2023

  6. [6]

    Gtac-gripper: A reconfigurable under-actuated four-fingered robotic gripper with tactile sensing,

    Z. Lu, H. Guo, W. Zhang, and H. Yu, “Gtac-gripper: A reconfigurable under-actuated four-fingered robotic gripper with tactile sensing,” IEEE Robot. Autom. Lett., vol. 7, no. 3, pp. 7232–7239, 2022

    2022

  7. [7]

    Gelsight: High-resolution robot tactile sensors for estimating geometry and force,

    W. Yuan, S. Dong, and E. H. Adelson, “Gelsight: High-resolution robot tactile sensors for estimating geometry and force,”Sensors, vol. 17, no. 12, p. 2762, 2017

    2017

  8. [8]

    Gelroller: A rolling vision-based tactile sensor for large surface reconstruction using self- supervised photometric stereo method,

    Z. Zhang, H. Ma, Y . Zhou, J. Ji, and H. Yang, “Gelroller: A rolling vision-based tactile sensor for large surface reconstruction using self- supervised photometric stereo method,” inProc. IEEE Int. Conf. Robot. Autom.IEEE, 2024, pp. 7961–7967

    2024

  9. [9]

    Inter-finger small object manipulation with densetact optical tactile sensor,

    W. K. Do, B. Aumann, C. Chungyoun, and M. Kennedy, “Inter-finger small object manipulation with densetact optical tactile sensor,”IEEE Robot. Autom. Lett., vol. 9, no. 1, pp. 515–522, 2023

    2023

  10. [10]

    Vibrissae-inspired vision- based magnetic-actuated whisker,

    Z. Hu, Y . Cheng, J. Wachs, and Y . She, “Vibrissae-inspired vision- based magnetic-actuated whisker,”Nat. Commun., vol. 17, no. 1, p. 939, 2025

    2025

  11. [11]

    Fruittouch: A perceptive gripper for gentle and scalable fruit harvesting,

    R. Zhang, M. A. Mirzaee, and W. Yuan, “Fruittouch: A perceptive gripper for gentle and scalable fruit harvesting,”IEEE Robot. Autom. Lett., vol. 11, no. 7, pp. 8276–8283, 2026

    2026

  12. [12]

    Rotipbot: Robotic handling of thin and flexible objects using rotatable tactile sensors,

    J. Jiang, X. Zhang, D. F. Gomes, T.-T. Do, and S. Luo, “Rotipbot: Robotic handling of thin and flexible objects using rotatable tactile sensors,”IEEE Trans. Robot., vol. 41, pp. 3684–3702, 2025

    2025

  13. [13]

    Cable manipulation with a tactile-reactive gripper,

    Y . She, S. Wang, S. Dong, N. Sunil, A. Rodriguez, and E. Adelson, “Cable manipulation with a tactile-reactive gripper,”Int. J. Robot. Res., vol. 40, no. 12-14, pp. 1385–1401, 2021

    2021

  14. [14]

    In-hand singulation, scooping, and cable untangling with a 5-dof tactile-reactive gripper,

    Y . Zhou, P. Zhou, S. Wang, and Y . She, “In-hand singulation, scooping, and cable untangling with a 5-dof tactile-reactive gripper,”Adv. Robot. Res., vol. 1, no. 2, p. 202500020, 2025

    2025

  15. [15]

    An active palm enhances dexterity of soft robotic in-hand manipulation,

    C. B. Teeple, G. R. Kim, M. A. Graule, and R. J. Wood, “An active palm enhances dexterity of soft robotic in-hand manipulation,” inProc. IEEE Int. Conf. Robot. Autom.IEEE, 2021, pp. 11 790–11 796

    2021

  16. [16]

    A soft robotic gripper with an active palm and reconfigurable fingers for fully dexterous in-hand manipulation,

    A. Pagoli, F. Chapelle, J. A. Corrales, Y . Mezouar, and Y . Lapusta, “A soft robotic gripper with an active palm and reconfigurable fingers for fully dexterous in-hand manipulation,”IEEE Robot. Autom. Lett., vol. 6, no. 4, pp. 7706–7713, 2021

    2021

  17. [17]

    Tactile-reactive gripper with an active palm for dexterous manipulation,

    Y . Zhou, W. S. Lee, Y . Gu, and Y . She, “Tactile-reactive gripper with an active palm for dexterous manipulation,”npj Robotics, vol. 4, p. 13, 2026

    2026

  18. [18]

    Soft robotic hand with tactile palm-finger coordination,

    N. Zhang, J. Ren, Y . Dong, X. Yang, R. Bian, J. Li, G. Gu, and X. Zhu, “Soft robotic hand with tactile palm-finger coordination,”Nat. Commun., vol. 16, no. 1, p. 2395, 2025

    2025

  19. [19]

    Look-to-touch: A vision-enhanced proximity and tactile sensor for distance and geometry perception in robotic manipulation,

    Y . Dong, J. Ren, Z. Liu, Z. Peng, Z. Yuan, N. Zhang, and G. Gu, “Look-to-touch: A vision-enhanced proximity and tactile sensor for distance and geometry perception in robotic manipulation,” IEEE/ASME Trans. Mechatron., 2026

    2026

  20. [20]

    Hy- dropalm: Dual-mode visual-tactile sensing for underwater humanoid robot hands,

    J. Ma, S. Cui, H. Chu, M. Tan, S. Wang, and Y . Wang, “Hy- dropalm: Dual-mode visual-tactile sensing for underwater humanoid robot hands,”IEEE Trans. Autom. Sci. Eng., vol. 23, pp. 3998–4008, 2026

    2026

  21. [21]

    Deep imitation learning for complex manipulation tasks from virtual reality teleoperation,

    T. Zhang, Z. McCarthy, O. Jow, D. Lee, X. Chen, K. Goldberg, and P. Abbeel, “Deep imitation learning for complex manipulation tasks from virtual reality teleoperation,” inProc. IEEE Int. Conf. Robot. Autom., 2018, pp. 5628–5635

    2018

  22. [22]

    Universal manipulation interface: In-the-wild robot teaching without in-the-wild robots,

    C. Chi, Z. Xu, C. Pan, E. Cousineau, B. Burchfiel, S. Feng, R. Tedrake, and S. Song, “Universal manipulation interface: In-the-wild robot teaching without in-the-wild robots,” inRobot.: Sci. Syst., 2024

    2024

  23. [23]

    Gello: A general, low- cost, and intuitive teleoperation framework for robot manipulators,

    P. Wu, Y . Shentu, Z. Yi, X. Lin, and P. Abbeel, “Gello: A general, low- cost, and intuitive teleoperation framework for robot manipulators,” inProc. IEEE/RSJ Int. Conf. Intell. Robots Syst.IEEE, 2024, pp. 12 156–12 163

    2024

  24. [24]

    Mobile aloha: Learning bimanual mobile manipulation using low-cost whole-body teleoperation,

    Z. Fu, T. Z. Zhao, and C. Finn, “Mobile aloha: Learning bimanual mobile manipulation using low-cost whole-body teleoperation,” in Proc. Conf. Robot Learn., 2024

    2024

  25. [25]

    Dexcap: Scalable and portable mocap data collection system for dexterous manipulation,

    C. Wang, H. Shi, W. Wang, R. Zhang, L. Fei-Fei, and C. K. Liu, “Dexcap: Scalable and portable mocap data collection system for dexterous manipulation,” inRobot.: Sci. Syst., 2024

    2024

  26. [26]

    Analyzing key objectives in human-to-robot retargeting for dexterous manipulation,

    C. Xin, M. Yu, Y . Jiang, Z. Zhang, and X. Li, “Analyzing key objectives in human-to-robot retargeting for dexterous manipulation,” IEEE Robot. Autom. Pract., vol. 1, pp. 29–34, 2026

    2026

  27. [27]

    Vt-refine: Learning bimanual assembly with visuo-tactile feedback via simulation fine-tuning,

    B. Huang, J. Xu, I. Akinola, W. Yang, B. Sundaralingam, R. O’Flaherty, D. Fox, X. Wang, A. Mousavian, Y .-W. Chaoet al., “Vt-refine: Learning bimanual assembly with visuo-tactile feedback via simulation fine-tuning,” inProc. Conf. Robot Learn., 2025

    2025

  28. [28]

    Flexitac: A low-cost, open-source, scal- able tactile sensing solution for robotic systems,

    B. Huang and Y . Li, “Flexitac: A low-cost, open-source, scal- able tactile sensing solution for robotic systems,”arXiv preprint arXiv:2604.28156, 2026

    Pith/arXiv arXiv 2026

  29. [29]

    Dexmv: Imitation learning for dexterous manipulation from human videos,

    Y . Qin, Y .-H. Wu, S. Liu, H. Jiang, R. Yang, Y . Fu, and X. Wang, “Dexmv: Imitation learning for dexterous manipulation from human videos,” inEuropean Conference on Computer Vision. Springer, 2022, pp. 570–587

    2022

  30. [30]

    In-hand singulation and scooping manipulation with a 5 dof tactile gripper,

    Y . Zhou, P. Zhou, S. Wang, and Y . She, “In-hand singulation and scooping manipulation with a 5 dof tactile gripper,” inProc. IEEE/RSJ Int. Conf. Intell. Robots Syst.IEEE, 2024, pp. 5238–5243

    2024

  31. [31]

    The ycb object and model set: Towards common benchmarks for manipulation research,

    B. Calli, A. Singh, A. Walsman, S. Srinivasa, P. Abbeel, and A. M. Dollar, “The ycb object and model set: Towards common benchmarks for manipulation research,” inProc. Int. Conf. Adv. Robot.IEEE, 2015, pp. 510–517

    2015

  32. [32]

    Manifeel: Benchmarking and understanding visuotactile manipulation policy learning,

    Q. K. Luu, P. Zhou, Z. Xu, Z. Zhang, Q. Qiu, and Y . She, “Manifeel: Benchmarking and understanding visuotactile manipulation policy learning,”arXiv preprint arXiv:2505.18472, 2025

    arXiv 2025

  33. [33]

    Dih-tele: Dexterous in-hand teleoperation framework for learning multiobjects manipulation with tactile sensing,

    J. Huang, K. Chen, J. Zhou, X. Lin, P. Abbeel, Q. Dou, and Y . Liu, “Dih-tele: Dexterous in-hand teleoperation framework for learning multiobjects manipulation with tactile sensing,”IEEE/ASME Trans. Mechatron., vol. 30, no. 5, pp. 3840–3851, 2025

    2025