pith. machine review for the scientific record. sign in

arxiv: 2605.04649 · v1 · submitted 2026-05-06 · 💻 cs.RO

Recognition: unknown

From Reach to Insert: Tactile-Augmented Precision Assembly under Sub-Millimeter Tolerances

Xinpan Meng , Siyao Huang , Jingpu Yang , Muyuan Ma , Zhenghua Ma , Lijun Han , Gao Yuan , Houcheng Li
show 1 more author
Long Cheng
Authors on Pith no claims yet

Pith reviewed 2026-05-08 15:57 UTC · model grok-4.3

classification 💻 cs.RO
keywords tactile sensingpeg-in-hole insertionprecision assemblyimitation learningreinforcement learningsub-millimeter tolerancescontact-rich manipulation
0
0 comments X

The pith

A two-stage tactile method using imitation then reinforcement learning achieves 67% success on 0.05 mm clearance insertions while cutting peak forces by 60%.

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

The paper develops a two-stage policy for peg-in-hole insertion tasks that must succeed despite sub-millimeter gaps where small errors cause jamming. Imitation learning first produces a reaching policy that moves the grasped peg near the hole from varied start poses. Reinforcement learning then takes over for the contact phase, using new tactile sampling and critic components to improve decisions and recover from contact errors. Systematic tests across five hole shapes and three clearances show higher success rates and lower forces than baselines, especially at the tightest tolerance.

Core claim

The authors show that a reaching policy trained by imitation learning followed by a reinforcement learning insertion policy, enhanced by tactile group sampling to cover key contact states and a tactile critic for more accurate value estimates, yields reliable insertions at clearances down to 0.05 mm with 67% success while lowering maximum interaction force by 60% and torque by 44%.

What carries the argument

Two-stage IL-to-RL pipeline with tactile group sampling for contact coverage and a tactile critic for policy evaluation.

If this is right

  • The RL insertion stage enables recovery from contact errors that would otherwise jam the assembly.
  • Success and force reductions hold across five hole geometries and three clearance settings.
  • Maximum interaction force drops 60% and torque 44% at the 0.05 mm clearance while success reaches 67%.
  • The separation into reaching and insertion phases keeps contact forces low enough for safe physical deployment.

Where Pith is reading between the lines

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

  • The same two-stage structure could be applied to other contact-rich tasks such as threading or connector mating.
  • Factories might use the method to reduce dependence on extremely precise vision systems for initial alignment.
  • Testing transfer across different robot arms or sensor placements would show how far the policies generalize.
  • Only the insertion stage would need retraining when switching to new part geometries.

Load-bearing premise

The tactile sampling and critic improvements transfer from simulation or limited training to real hardware across five hole geometries and three clearances without per-task retuning or overfitting.

What would settle it

A real-robot test on the five hole geometries at 0.05 mm clearance that yields success below 50% or force reduction below 30% would show the claimed generalization and safety gains do not hold.

Figures

Figures reproduced from arXiv: 2605.04649 by Gao Yuan, Houcheng Li, Jingpu Yang, Lijun Han, Long Cheng, Muyuan Ma, Siyao Huang, Xinpan Meng, Zhenghua Ma.

Figure 1
Figure 1. Figure 1: Overview of the proposed two-stage tactile-guided assembly view at source ↗
Figure 2
Figure 2. Figure 2: Overview of prior peg-in-hole works categorized by task setting view at source ↗
Figure 3
Figure 3. Figure 3: Overview of the proposed two-stage framework for peg-in-hole insertion. In the view at source ↗
Figure 4
Figure 4. Figure 4: Illustration of the tactile baseline estimation and the resulting tactile view at source ↗
Figure 7
Figure 7. Figure 7: Aggregated Critic loss during online reinforcement learning. The view at source ↗
Figure 8
Figure 8. Figure 8: Comparison of Q-value disagreement during training. The proposed view at source ↗
Figure 9
Figure 9. Figure 9: Interaction force and torque comparison results between vanilla IL+RL and the proposed method during real-world 0.05 mm insertion. Vanilla view at source ↗
read the original abstract

High-precision assembly frequently involves tight-tolerance insertions, where even slight pose errors can cause jamming or excessive interaction forces, making robust and safe insertion policies difficult to obtain. This paper proposes a tactile-augmented two-stage method that combines Imitation Learning (IL) and Reinforcement Learning (RL) for precision insertion tasks. In the first stage, IL learns a reaching policy with position generalization that grasps the peg and brings it to the vicinity of the target region. In the second stage, RL executes the insertion and enables recovery from failures during contact-rich interactions. To better exploit tactile feedback, we introduce tactile group sampling to increase coverage of critical contact segments during training, and design a tactile critic to more accurately evaluate policy values, improving insertion performance while maintaining low contact forces. We conduct systematic experiments across five hole geometries and three clearance settings. Results show that our method substantially improves insertion performance across all settings; under the most challenging 0.05\,mm clearance, it achieves a 67\% success rate while keeping contact forces low, reducing the maximum interaction force by 60\% and torque by 44\%, thereby validating both effectiveness and safety for precision assembly.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

2 major / 1 minor

Summary. The manuscript proposes a tactile-augmented two-stage pipeline that first uses imitation learning to learn a reaching policy with position generalization and then applies reinforcement learning for the insertion phase, augmented by tactile group sampling to improve contact coverage and a tactile critic for better value estimation. Systematic experiments are reported across five hole geometries and three clearance values (including 0.05 mm), claiming a 67% success rate at the tightest clearance together with 60% and 44% reductions in maximum interaction force and torque, respectively.

Significance. If the empirical gains are shown to be robust and generalizable, the work would offer a practical route to safer contact-rich insertion under sub-millimeter tolerances without requiring ultra-precise hardware, which is relevant to industrial assembly robotics.

major comments (2)
  1. [Results] Results section (and abstract): The headline quantitative claims (67% success at 0.05 mm clearance, 60% force reduction, 44% torque reduction) are presented without baseline comparisons, number of trials, statistical tests, error bars, or explicit definitions of how success and force/torque metrics were computed and aggregated. These omissions prevent assessment of whether the reported improvements are attributable to the tactile group sampling and tactile critic or to other factors.
  2. [Methods] Methods section: The description of the two-stage IL+RL pipeline and the two tactile-specific components leaves open whether the tactile group sampling and tactile critic were trained once (in simulation or on a subset of geometries) and then deployed zero-shot across all five hole geometries and three clearances on physical hardware. Given the sensitivity of tactile signals to calibration drift, surface compliance, and friction, any hidden per-geometry retuning would render the cross-setting generalization claim non-falsifiable on the basis of the reported evidence.
minor comments (1)
  1. [Abstract] Abstract: The phrase 'systematic experiments across five hole geometries and three clearance settings' should be accompanied by the total number of trials and the exact clearance values to allow immediate evaluation of the scope.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the constructive comments, which have helped us improve the clarity and rigor of the manuscript. We address each major comment below and have revised the manuscript to incorporate the requested details and clarifications.

read point-by-point responses

  1. Referee: [Results] Results section (and abstract): The headline quantitative claims (67% success at 0.05 mm clearance, 60% force reduction, 44% torque reduction) are presented without baseline comparisons, number of trials, statistical tests, error bars, or explicit definitions of how success and force/torque metrics were computed and aggregated. These omissions prevent assessment of whether the reported improvements are attributable to the tactile group sampling and tactile critic or to other factors.

    Authors: We agree that the original presentation omitted key experimental details. In the revised manuscript we have expanded the Results section (and updated the abstract) to include: baseline comparisons against pure IL, pure RL, and non-tactile variants; the number of trials (50 independent rollouts per condition); error bars showing standard deviation; and statistical tests (paired t-tests, p < 0.01 for the reported gains). Success is now explicitly defined as full insertion to target depth within the time limit without jamming or force-limit violation; force and torque metrics are the peak values recorded during the contact phase, averaged only over successful trials. These additions confirm that the performance and safety improvements are attributable to the tactile group sampling and tactile critic. revision: yes

  2. Referee: [Methods] Methods section: The description of the two-stage IL+RL pipeline and the two tactile-specific components leaves open whether the tactile group sampling and tactile critic were trained once (in simulation or on a subset of geometries) and then deployed zero-shot across all five hole geometries and three clearances on physical hardware. Given the sensitivity of tactile signals to calibration drift, surface compliance, and friction, any hidden per-geometry retuning would render the cross-setting generalization claim non-falsifiable on the basis of the reported evidence.

    Authors: The tactile group sampling and tactile critic were trained once in simulation on a distribution that randomized hole geometries and clearances; the resulting policies were then deployed zero-shot on the physical robot for all five test geometries and three clearances with no per-geometry retuning, fine-tuning, or hardware-specific adjustments. We have revised the Methods section to state this procedure explicitly, including the simulation training protocol and the sim-to-real transfer steps taken to handle tactile signal variability. This makes the zero-shot generalization claim directly verifiable from the reported evidence. revision: yes

Circularity Check

0 steps flagged

No circularity: purely empirical validation of IL+RL pipeline

full rationale

The paper describes a two-stage IL-then-RL method augmented with tactile group sampling and a tactile critic, then reports success rates and force reductions from physical experiments across five geometries and three clearances. No equations, derivations, fitted parameters, or uniqueness theorems appear in the provided text. Performance numbers are direct experimental outcomes, not quantities defined in terms of themselves or obtained by renaming a self-citation. The method is therefore self-contained against external benchmarks.

Axiom & Free-Parameter Ledger

0 free parameters · 0 axioms · 0 invented entities

Abstract-only review yields no explicit free parameters, axioms, or invented entities; the approach relies on standard assumptions of IL/RL applicability to robotic insertion and the utility of tactile signals for contact-rich control.

pith-pipeline@v0.9.0 · 5533 in / 1064 out tokens · 27603 ms · 2026-05-08T15:57:23.735004+00:00 · methodology

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

40 extracted references · 6 canonical work pages · 1 internal anchor

  1. [1]

    Realtime State Estimation with Tactile and Visual Sensing. Application to Planar Manipulation,

    K.-T. Yu and A. Rodriguez, “Realtime State Estimation with Tactile and Visual Sensing. Application to Planar Manipulation,” inIEEE Int. Conf. on Robotics and Automation (ICRA), 2018

    2018

  2. [2]

    Learning to Fill the Seam by Vision: Sub-millimeter Peg-in- hole on Unseen Shapes in Real World,

    L. Xie, H. Yu, Y . Zhao, H. Zhang, Z. Zhou, M. Wang, Y . Wang, and R. Xiong, “Learning to Fill the Seam by Vision: Sub-millimeter Peg-in- hole on Unseen Shapes in Real World,” inIEEE Int. Conf. on Robotics and Automation (ICRA), 2022

    2022

  3. [3]

    AutoMate: Specialist and Generalist Assembly Policies over Diverse Geometries,

    B. Tang, I. Akinola, J. Xu, B. Wen, A. Handa, K. Van Wyk, D. Fox, G. S. Sukhatme, F. Ramos, and Y . Narang, “AutoMate: Specialist and Generalist Assembly Policies over Diverse Geometries,” inProc. Robotics: Science and Systems (RSS), 2024

    2024

  4. [4]

    Feedback Deep Deterministic Policy Gradient With Fuzzy Reward for Robotic Multiple Peg-in-Hole Assembly Tasks,

    J. Xu, Z. Hou, W. Wang, B. Xu, K. Zhang, and K. Chen, “Feedback Deep Deterministic Policy Gradient With Fuzzy Reward for Robotic Multiple Peg-in-Hole Assembly Tasks,”IEEE Trans. Ind. Informat., 2019

    2019

  5. [5]

    Interpretation of force and moment signals for compliant peg-in-hole assembly,

    W. S. Newman, Y . Zhao, and Y .-H. Pao, “Interpretation of force and moment signals for compliant peg-in-hole assembly,” inIEEE Int. Conf. on Robotics and Automation (ICRA), 2001

    2001

  6. [6]

    Learning Variable Impedance Control via Inverse Reinforcement Learning for Force- Related Tasks,

    X. Zhang, L. Sun, Z. Kuang, and M. Tomizuka, “Learning Variable Impedance Control via Inverse Reinforcement Learning for Force- Related Tasks,”IEEE Robot. Autom. Lett., 2021

    2021

  7. [7]

    Robust Peg-in-Hole Assembly under Uncertainties via Compliant and Interactive Contact-Rich Manipulation,

    Y . Chen, K. Kimble, H. H. Qian, P. Chanrungmaneekul, R. Seney, and K. Hang, “Robust Peg-in-Hole Assembly under Uncertainties via Compliant and Interactive Contact-Rich Manipulation,” inProc. Robotics: Science and Systems (RSS), 2025

    2025

  8. [8]

    Surprisingly Robust In-Hand Manipulation: An Empirical Study,

    A. Bhatt, A. Sieler, S. Puhlmann, and O. Brock, “Surprisingly Robust In-Hand Manipulation: An Empirical Study,” inProc. Robotics: Science and Systems (RSS), 2021

    2021

  9. [9]

    Compliant Peg- in-Hole Assembly Using Partial Spiral Force Trajectory With Tilted Peg Posture,

    H. Park, J. Park, D.-H. Lee, J.-H. Park, and J.-H. Bae, “Compliant Peg- in-Hole Assembly Using Partial Spiral Force Trajectory With Tilted Peg Posture,”IEEE Robot. Autom. Lett., 2020

    2020

  10. [10]

    1 kHz Behavior Tree for Self-adaptable Tactile Insertion,

    Y . Wu, F. Wu, L. Chen, K. Chen, S. Schneider, L. Johannsmeier, Z. Bing, F. J. Abu-Dakka, A. Knoll, and S. Haddadin, “1 kHz Behavior Tree for Self-adaptable Tactile Insertion,” inIEEE Int. Conf. on Robotics and Automation (ICRA), 2024

    2024

  11. [11]

    Active Extrinsic Contact Sensing: Applica- tion to General Peg-in-Hole Insertion,

    S. Kim and A. Rodriguez, “Active Extrinsic Contact Sensing: Applica- tion to General Peg-in-Hole Insertion,” inIEEE Int. Conf. on Robotics and Automation (ICRA), 2022

    2022

  12. [12]

    Quickly Inserting Pegs into Uncertain Holes using Multi-view Images and Deep Network Trained on Synthetic Data,

    J. C. Triyonoputro, W. Wan, and K. Harada, “Quickly Inserting Pegs into Uncertain Holes using Multi-view Images and Deep Network Trained on Synthetic Data,” inIEEE/RSJ Int. Conf. on Intelligent Robots and Systems (IROS), 2019

    2019

  13. [13]

    Learning to Assemble: Estimating 6D Poses for Robotic Object-Object Manipulation,

    S. Stevsic, S. Christen, and O. Hilliges, “Learning to Assemble: Estimating 6D Poses for Robotic Object-Object Manipulation,”IEEE Robot. Autom. Lett., 2020

    2020

  14. [14]

    Understanding Multi- modal Perception Using Behavioral Cloning for Peg-in-a-Hole Insertion Tasks,

    Y . Liu, D. Romeres, D. K. Jha, and D. Nikovski, “Understanding Multi- modal Perception Using Behavioral Cloning for Peg-in-a-Hole Insertion Tasks,” inProc. RSS Workshop on Advances & Challenges in Imitation Learning for Robotics, 2020

    2020

  15. [15]

    Contact-Rich Robotic Assembly in Construction via Diffusion Policy Learning

    S. Mozaffari, D. Ruan, W. van den Bogert, N. Fazeli, S. Adriaenssens, and A. Adel, “Learning Diffusion Policies for Robotic Manipulation of Timber Joinery under Fabrication Uncertainty,”arXiv preprint arXiv:2511.17774, 2025

    work page internal anchor Pith review Pith/arXiv arXiv 2025

  16. [16]

    Touch begins where vision ends: Generalizable policies for contact-rich manipula- tion,

    Z. Zhao, S. Haldar, J. Cui, L. Pinto, and R. Bhirangi, “Touch begins where vision ends: Generalizable policies for contact-rich manipula- tion,”arXiv preprint arXiv:2506.13762, 2025

    work page arXiv 2025

  17. [17]

    ManipLLM: Embodied Multimodal Large Language Model for Object-Centric Robotic Manipulation,

    X. Li, M. Zhang, Y . Geng, H. Geng, Y . Long, Y . Shen, R. Zhang, J. Liu, and H. Dong, “ManipLLM: Embodied Multimodal Large Language Model for Object-Centric Robotic Manipulation,” inProc. IEEE/CVF Conf. on Computer Vision and Pattern Recognition (CVPR), 2024

    2024

  18. [18]

    Making Sense of Vision and Touch: Self- Supervised Learning of Multimodal Representations for Contact-Rich Tasks,

    M. A. Lee, Y . Zhu, K. Srinivasan, P. Shah, S. Savarese, L. Fei-Fei, A. Garg, and J. Bohg, “Making Sense of Vision and Touch: Self- Supervised Learning of Multimodal Representations for Contact-Rich Tasks,” inIEEE Int. Conf. on Robotics and Automation (ICRA), 2019

    2019

  19. [19]

    Reinforcement Learning on Variable Impedance Controller for High-Precision Robotic Assembly,

    J. Luo, E. Solowjow, C. Wen, J. A. Ojea, A. M. Agogino, A. Tamar, and P. Abbeel, “Reinforcement Learning on Variable Impedance Controller for High-Precision Robotic Assembly,” inIEEE Int. Conf. on Robotics and Automation (ICRA), 2019

    2019

  20. [20]

    Meta-reinforcement Learning for Robotic Industrial Insertion Tasks,

    G. Schoettler, A. Nair, J. A. Ojea, S. Levine, and E. Solowjow, “Meta-reinforcement Learning for Robotic Industrial Insertion Tasks,” inIEEE/RSJ Int. Conf. on Intelligent Robots and Systems (IROS), 2020

    2020

  21. [21]

    Offline Meta-Reinforcement Learning for Industrial Insertion,

    T. Z. Zhao, J. Luo, O. Sushkov, R. Pevceviciute, N. Heess, J. Scholz, S. Schaal, and S. Levine, “Offline Meta-Reinforcement Learning for Industrial Insertion,” inIEEE Int. Conf. on Robotics and Automation (ICRA), 2022

    2022

  22. [22]

    FORGE: Force-Guided Exploration for Robust Contact-Rich Manipulation under Uncertainty,

    M. Noseworthy, B. Tang, B. Wen, A. Handa, C. Kessens, N. Roy, D. Fox, F. Ramos, Y . Narang, and I. Akinola, “FORGE: Force-Guided Exploration for Robust Contact-Rich Manipulation under Uncertainty,” IEEE Robot. Autom. Lett., 2025

    2025

  23. [23]

    IndustReal: Transferring Contact-Rich Assembly Tasks from Simulation to Reality,

    B. Tang, M. A. Lin, I. Akinola, A. Handa, G. S. Sukhatme, F. Ramos, D. Fox, and Y . Narang, “IndustReal: Transferring Contact-Rich Assembly Tasks from Simulation to Reality,” inProc. Robotics: Science and Systems (RSS), 2023

    2023

  24. [24]

    Precise and Dexterous Robotic Manipulation via Human-In-The-Loop Reinforcement Learning,

    J. Luo, C. Xu, J. Wu, and S. Levine, “Precise and Dexterous Robotic Manipulation via Human-In-The-Loop Reinforcement Learning,”Sci- ence Robotics, 2025

    2025

  25. [25]

    Efficient Online Reinforcement Learning with Offline Data,

    P. J. Ball, L. Smith, I. Kostrikov, and S. Levine, “Efficient Online Reinforcement Learning with Offline Data,” inInt. Conf. on Machine Learning (ICML), 2023

    2023

  26. [26]

    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. Chao, and Y . Li, “VT- Refine: Learning Bimanual Assembly with Visuo-Tactile Feedback via Simulation Fine-Tuning,” inProc. Conf. on Robot Learning (CoRL), 2025

    2025

  27. [27]

    Residual Reinforcement Learning for Robot Control,

    T. Johannink, S. Bahl, A. Nair, J. Luo, A. Kumar, M. Loskyll, J. A. Ojea, E. Solowjow, and S. Levine, “Residual Reinforcement Learning for Robot Control,” inIEEE Int. Conf. on Robotics and Automation (ICRA), 2019

    2019

  28. [28]

    Deep Reinforcement Learning for High Precision Assembly Tasks,

    T. Inoue, G. De Magistris, A. Munawar, T. Yokoya, and R. Tachibana, “Deep Reinforcement Learning for High Precision Assembly Tasks,” in IEEE/RSJ Int. Conf. on Intelligent Robots and Systems (IROS), 2017

    2017

  29. [29]

    Tactile-RL for Insertion: Generalization to Objects of Unknown Geometry,

    S. Dong, D. K. Jha, D. Romeres, S. Kim, D. Nikovski, and A. Rodriguez, “Tactile-RL for Insertion: Generalization to Objects of Unknown Geometry,” inIEEE Int. Conf. on Robotics and Automation (ICRA), 2021

    2021

  30. [30]

    A Tactile-Proximity Dual-Mode Photoelectric Sensor: Implementation and Applications,

    X. Meng, L. Cheng, and Z. Li, “A Tactile-Proximity Dual-Mode Photoelectric Sensor: Implementation and Applications,”IEEE Trans. Robot., 2025

    2025

  31. [31]

    Tactile-based Active Inference for Force-Controlled Peg-in-Hole In- sertions,

    T. Kamijo, I. G. Ramirez-Alpizar, E. Coronado, and G. Venture, “Tactile-based Active Inference for Force-Controlled Peg-in-Hole In- sertions,”arXiv preprint arXiv:2309.15681, 2023

    work page arXiv 2023

  32. [32]

    Reactive Diffusion Policy: Slow-Fast Visual-Tactile Policy Learning for Contact-Rich Manipulation,

    H. Xue, J. Ren, W. Chen, G. Zhang, Y . Fang, G. Gu, H. Xu, and C. Lu, “Reactive Diffusion Policy: Slow-Fast Visual-Tactile Policy Learning for Contact-Rich Manipulation,” inProc. Robotics: Science and Systems (RSS), 2025

    2025

  33. [33]

    TacDiffusion: Force-domain Diffusion Policy for Precise Tactile Manipulation,

    Y . Wu, Z. Chen, F. Wu, L. Chen, L. Zhang, Z. Bing, A. Swikir, S. Haddadin, and A. Knoll, “TacDiffusion: Force-domain Diffusion Policy for Precise Tactile Manipulation,” inIEEE Int. Conf. on Robotics and Automation (ICRA), 2025

    2025

  34. [34]

    Vla-touch: Enhancing vision- language-action models with dual-level tactile feedback

    J. Bi, K. Y . Ma, C. Hao, M. Z. Shou, and H. Soh, “VLA-Touch: Enhancing Vision-Language-Action Models with Dual-Level Tactile Feedback,”arXiv preprint arXiv:2507.17294, 2025

    work page arXiv 2025

  35. [35]

    MimicTouch: Leveraging Multi-modal Human Tactile Demonstrations for Contact- rich Manipulation,

    K. Yu, Y . Han, Q. Wang, V . Saxena, D. Xu, and Y . Zhao, “MimicTouch: Leveraging Multi-modal Human Tactile Demonstrations for Contact- rich Manipulation,” inProc. Conf. on Robot Learning (CoRL), 2024

    2024

  36. [36]

    Visuotactile-RL: Learning Multimodal Manipulation Policies with Deep Reinforcement Learning,

    J. Hansen, F. Hogan, D. Rivkin, D. Meger, M. Jenkin, and G. Dudek, “Visuotactile-RL: Learning Multimodal Manipulation Policies with Deep Reinforcement Learning,” inIEEE Int. Conf. on Robotics and Automation (ICRA), 2022

    2022

  37. [37]

    Visuotactile-Based Learning for Insertion with Compliant Hands,

    O. Azulay, D. M. Ramesh, N. Curtis, and A. Sintov, “Visuotactile-Based Learning for Insertion with Compliant Hands,”IEEE Robot. Autom. Lett., 2025

    2025

  38. [38]

    Tla: Tactile-language-action model for contact-rich manipu- lation.arXiv preprint arXiv:2503.08548, 2025

    P. Hao, C. Zhang, D. Li, X. Cao, X. Hao, S. Cui, and S. Wang, “TLA: Tactile-Language-Action Model for Contact-Rich Manipulation,”arXiv preprint arXiv:2503.08548, 2025

    work page arXiv 2025

  39. [39]

    Enhancing Tactile-based Reinforcement Learning for Robotic Con- trol,

    E. Miller, T. McInroe, D. Abel, O. Mac Aodha, and S. Vijayakumar, “Enhancing Tactile-based Reinforcement Learning for Robotic Con- trol,” inAdvances in Neural Information Processing Systems (NeurIPS), 2025

    2025

  40. [40]

    Dexterity from touch: Self-supervised pre-training of tactile representations with robotic play.arXiv preprint arXiv:2303.12076, 2023

    I. Guzey, B. Evans, S. Chintala, and L. Pinto, “Dexterity from Touch: Self-Supervised Pre-Training of Tactile Representations with Robotic Play,”arXiv preprint arXiv:2303.12076, 2023

    work page arXiv 2023