pith. sign in

arxiv: 2606.20193 · v1 · pith:MIMLDFH6new · submitted 2026-06-18 · 💻 cs.RO

Belt-Finger: An Affordable Soft Belt-Driven Gripper for Dexterous In-Hand Manipulation

Boya Zhang , Andreas Zell , Georg Martius This is my paper

Pith reviewed 2026-06-26 17:09 UTC · model grok-4.3

classification 💻 cs.RO
keywords soft roboticsparallel gripperin-hand manipulationbelt-driven mechanismdexterous manipulationteleoperationmodel predictive control
0
0 comments X

The pith

A double-soft-belt finger module adds three in-hand degrees of freedom to parallel grippers while keeping them simple and cheap.

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

The paper presents an upgrade to standard parallel-jaw grippers using a double-soft-belt finger module. This module keeps the basic open-close function but adds three new in-hand motions: sliding along the finger, tilting, and rotating. The design focuses on being cheap to build and easy to attach to existing robots. By testing it with different control methods on tricky tasks, the authors show it makes manipulation more capable than using a regular gripper alone.

Core claim

The paper claims that replacing or upgrading parallel gripper fingers with a double-soft-belt mechanism allows for in-hand translation, pitch, and roll while maintaining the original gripper's simplicity and precision. This is achieved through a deliberately simple design suited for inexpensive manufacturing. The added capabilities are shown to enhance dexterity in tasks using model predictive control, teleoperation, and trained policies, making previously difficult manipulations feasible without changing the arm or adding complexity.

What carries the argument

double-soft-belt-based finger module that adds translation, pitch, and roll to the standard parallel gripper motion

If this is right

  • The gripper enables manipulation in confined spaces by reducing reliance on arm motion.
  • It integrates with existing control methods like MPC for known objects.
  • Teleoperation becomes more effective with simultaneous arm and gripper control.
  • Task success rates increase compared to conventional grippers across various control strategies.

Where Pith is reading between the lines

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

  • This approach might allow smaller robots to perform tasks that currently require larger workspaces.
  • Researchers could adapt the belt design for other end-effectors to add degrees of freedom affordably.
  • The low cost could make advanced manipulation accessible to more educational and hobbyist setups.

Load-bearing premise

The belt mechanism can be manufactured inexpensively and integrated while preserving the reliability and precise control of traditional parallel grippers.

What would settle it

A direct comparison experiment showing that the belt-finger gripper fails to complete any of the manipulation tasks that the conventional gripper can perform, or exhibits frequent mechanical failures during use.

Figures

Figures reproduced from arXiv: 2606.20193 by Andreas Zell, Boya Zhang, Georg Martius.

Figure 1
Figure 1. Figure 1: Belt-Finger design. The left side shows the gripper made of two belt-enhanced jaws grasping a dummy object. Three colors show the in-hand manipulation DoFs in the local frame. The right side shows the exploded view of the Belt-Finger. There are three active wheels, each driven by a servo motor, with a coating layer outside of each belt to increase friction forces. 1 INTRODUCTION The applicability and succe… view at source ↗
Figure 2
Figure 2. Figure 2: Comparison between the Belt￾Finger (top) and a close variant (bottom) on picking and reorientation. 3 Hardware Design The hardware comprises three components: finger, belt, and gripper base. 2 [PITH_FULL_IMAGE:figures/full_fig_p002_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: In-hand motion benchmarking with teleoperation. [PITH_FULL_IMAGE:figures/full_fig_p003_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: Model predictive control setup. The gripper model is shown in green, and the iCEM optimizer is shown in blue. l1, l2 and r1, r2 are left and right belts. C = c1−4 are approximated contact points, where green means active controlled, while red is passive. In iCEM, for each executed action, nopt iterations of optimization will be performed, starting with nsp samples. During each optimization iteration, traje… view at source ↗
Figure 5
Figure 5. Figure 5: The teleoperation setup for the Belt-Finger [PITH_FULL_IMAGE:figures/full_fig_p005_5.png] view at source ↗
Figure 6
Figure 6. Figure 6: Payload benchmarking setup. The in￾hand payloads of four shapes are measured. The three surface material are listed in [PITH_FULL_IMAGE:figures/full_fig_p006_6.png] view at source ↗
Figure 7
Figure 7. Figure 7: Comparison of MPC rollouts in simulation and real-world environments. Results demonstrate the MPC effectively reaches targets in easy and medium cases (the loss is notably optimized against orientation). The MPC minimizes errors even for theoretically unreachable targets in the combi-cases which violate the force-closure. Residual errors ( [PITH_FULL_IMAGE:figures/full_fig_p006_7.png] view at source ↗
Figure 9
Figure 9. Figure 9: Showcasing a card insertion task using the teleoperation setup. The card is picked up from the table and inserted horizontally in both (a) and (c), using different strategies. 6.3 Manipulation of daily objects We demonstrate the gripper’s teleoperated manipulation capabilities through two challenging tasks. Task One: Table Cleaning. The operator picks, reorients, and neatly places irregularly shaped object… view at source ↗
Figure 10
Figure 10. Figure 10: VLA control of the Belt-Finger. We evaluate five challenging daily tasks requiring complex in-hand manipulation. Finetuned VLA models successfully leverage the soft belts to achieve high success rates, demonstrating closed-loop corrections on tasks inaccessible to standard parallel grippers. completed three trials each. As shown in [PITH_FULL_IMAGE:figures/full_fig_p008_10.png] view at source ↗
Figure 11
Figure 11. Figure 11: User study. The success rate and total time used for finishing the pen insertion task across participants. The Belt-Finger has several limitations: (i) due to the double-belt layout the pitch for small objects is limited and the usable workspace for very thin/large/concave shapes is reduced; (ii) manipulatability depends on friction/adhesion of the coating and pressure distribution, making rotations nonli… view at source ↗
Figure 12
Figure 12. Figure 12: Comparison between the conventional parallel finger with the Belt-Finger on ten daily tasks. Task descriptions are listed in [PITH_FULL_IMAGE:figures/full_fig_p013_12.png] view at source ↗
Figure 13
Figure 13. Figure 13: In-hand motion for complex objects with MPC. [PITH_FULL_IMAGE:figures/full_fig_p014_13.png] view at source ↗
Figure 14
Figure 14. Figure 14: Rotatable transmission structure. This design released the rotation DoF around the z-axis. However, the extra motion free￾dom reduces the belt stability compared to the simpler design [PITH_FULL_IMAGE:figures/full_fig_p017_14.png] view at source ↗
Figure 15
Figure 15. Figure 15: Fixed wheel in the simpler design results in a less complete wrapping when the non-convex object has multiple contact areas along the belt [PITH_FULL_IMAGE:figures/full_fig_p017_15.png] view at source ↗
read the original abstract

Parallel-jaw grippers are the default manipulator choice in robotics because they are simple, robust, and inexpensive. Their limited in-hand mobility, however, often forces large arm motions and restricts dexterous manipulation in confined workspaces. We present a parallel-gripper upgrade: a double-soft-belt-based finger module that preserves standard opening/closing while adding three in-hand degrees of freedom (DoF): translation, pitch, and roll. The mechanism is deliberately kept simple and engineered for inexpensive manufacturing and straightforward integration, preserving the reliability and precise control of traditional parallel grippers while greatly broadening the range of manipulation capabilities. To demonstrate the utility of the added DoFs, we integrate the gripper in two control pipelines. First, we adapt a model predictive controller for in-hand manipulation of known objects. Second, we introduce a lightweight teleoperation interface that enables simultaneous control of the robot arm and gripper (10 DoFs total) with minimal hardware. Across a suite of challenging manipulation tasks executed via teleoperation, MPC, and trained policies, the proposed gripper consistently improves dexterity and task feasibility compared to a conventional parallel gripper

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 / 0 minor

Summary. The paper introduces Belt-Finger, a double-soft-belt finger module designed as an upgrade to standard parallel-jaw grippers. It preserves the original opening/closing motion while adding three in-hand DoFs (translation, pitch, roll) via a belt-driven mechanism. The design is presented as simple, inexpensive to manufacture, and capable of preserving the reliability and precise control of conventional parallel grippers. Utility is shown by integrating the gripper into teleoperation (10-DoF interface), an adapted MPC controller for known objects, and trained policies, with the claim that it consistently improves dexterity and task feasibility over a baseline parallel gripper across challenging manipulation tasks.

Significance. If the performance claims are supported by quantitative evidence, the work could provide a practical, low-cost path to add in-hand dexterity to the most common gripper type without sacrificing its core advantages, potentially enabling more manipulation in confined spaces. The absence of any reported metrics, baselines, or reliability data in the abstract, however, leaves the significance currently unassessable.

major comments (2)
  1. [Abstract] Abstract: the central claim that the gripper 'consistently improves dexterity and task feasibility compared to a conventional parallel gripper' across teleoperation, MPC, and policy tasks is presented without any quantitative results, error bars, baseline comparisons, task success rates, or experimental protocol. This directly undermines evaluation of the primary contribution.
  2. [Abstract] Abstract: the assertion that the belt mechanism 'preserves the reliability and precise control of traditional parallel grippers' while adding DoFs is load-bearing for the comparison but is unsupported by any specifications on belt tension, wear, slippage thresholds, position error under load, or failure-mode analysis. Without these, the feasibility gains cannot be shown to generalize beyond the specific tasks.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the constructive feedback. We address the two major comments on the abstract point by point below.

read point-by-point responses

  1. Referee: [Abstract] Abstract: the central claim that the gripper 'consistently improves dexterity and task feasibility compared to a conventional parallel gripper' across teleoperation, MPC, and policy tasks is presented without any quantitative results, error bars, baseline comparisons, task success rates, or experimental protocol. This directly undermines evaluation of the primary contribution.

    Authors: We agree that the abstract would benefit from quantitative support for the performance claims. In the revised manuscript we will update the abstract to include key metrics such as task success rates and baseline comparisons drawn from the experimental sections. revision: yes

  2. Referee: [Abstract] Abstract: the assertion that the belt mechanism 'preserves the reliability and precise control of traditional parallel grippers' while adding DoFs is load-bearing for the comparison but is unsupported by any specifications on belt tension, wear, slippage thresholds, position error under load, or failure-mode analysis. Without these, the feasibility gains cannot be shown to generalize beyond the specific tasks.

    Authors: The manuscript validates reliable operation through successful task execution without belt-related failures, but we acknowledge the abstract lacks explicit supporting specifications. We will revise the abstract to reference the design's demonstrated robustness and point to the detailed mechanism analysis in the body of the paper. revision: yes

Circularity Check

0 steps flagged

No circularity: hardware design and task results are self-contained

full rationale

The paper presents a mechanical gripper upgrade and validates it through task demonstrations using teleoperation, MPC, and policies. No equations, parameters, derivations, or self-citations appear in the provided text that could reduce a claimed result to its own inputs by construction. The central claims rest on direct experimental comparison rather than any load-bearing logical loop. This matches the expected non-circular outcome for a hardware-focused robotics paper.

Axiom & Free-Parameter Ledger

0 free parameters · 1 axioms · 1 invented entities

The central claim rests on the unverified premise that the belt-driven module can be built cheaply and reliably while adding functional DoFs; no free parameters or invented physical entities are introduced in the abstract.

axioms (1)
  • domain assumption Parallel-jaw grippers are the default manipulator choice because they are simple, robust, and inexpensive.
    Opening sentence of the abstract.
invented entities (1)
  • Double-soft-belt finger module no independent evidence
    purpose: To add translation, pitch, and roll DoFs to a parallel gripper.
    New hardware concept introduced to solve the stated limitation.

pith-pipeline@v0.9.1-grok · 5731 in / 1206 out tokens · 25896 ms · 2026-06-26T17:09:49.560525+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

31 extracted references · 8 canonical work pages · 2 internal anchors

  1. [1]

    URL https://shadowrobot.com/ dexterous-hand-series/

    Shadowrobot dexterous hand. URL https://shadowrobot.com/ dexterous-hand-series/. 2025

    2025

  2. [2]

    K. Shaw, A. Agarwal, and D. Pathak. Leap hand: Low-cost, efficient, and anthropomorphic hand for robot learning.arXiv preprint arXiv:2309.06440, 2023

    work page arXiv 2023

  3. [3]

    R. C. Ornelas, T. Cantú, I. Sperandio, A. H. Slocum, and P. Agrawal. Everyday finger: a robotic finger that meets the needs of everyday interactive manipulation. In2024 IEEE International Conference on Robotics and Automation (ICRA), pages 16016–16023. IEEE, 2024

    2024

  4. [4]

    T. Mack, K. Zhang, and K. Althoefer. A soft robotic gripper with active palm for in-hand object reorientation.arXiv preprint arXiv:2307.13657, 2023

    work page arXiv 2023

  5. [5]

    C. B. Teeple, R. C. S. Louis, M. A. Graule, and R. J. Wood. The role of digit arrangement in soft robotic in-hand manipulation. In2021 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), pages 7201–7208. IEEE, 2021

    2021

  6. [6]

    Jiang, X

    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 Transactions on Robotics, 2025

    2025

  7. [7]

    A. J. Spiers, B. Calli, and A. M. Dollar. Variable-friction finger surfaces to enable within-hand manipulation via gripping and sliding.IEEE Robotics and Automation Letters, 3(4):4116–4123, 2018

    2018

  8. [8]

    Tincani, G

    V . Tincani, G. Grioli, M. G. Catalano, M. Garabini, S. Grechi, G. Fantoni, and A. Bicchi. Implementation and control of the velvet fingers: a dexterous gripper with active surfaces. In 2013 IEEE International Conference on Robotics and Automation, pages 2744–2750. IEEE, 2013

    2013

  9. [9]

    Xiang, J

    S. Xiang, J. Li, Y . Zhang, Y . Yang, J. Liu, and Z. Liu. Adaptive wrapping with active elastic band-based gripper for stable in-hand manipulation.Sensors and Actuators A: Physical, 377: 115743, 2024

    2024

  10. [10]

    J. Xu, L. Wu, C. Lin, D. Zhao, and H. Xu. Dtactive: A vision-based tactile sensor with active surface.arXiv preprint arXiv:2410.08337, 2024

    work page arXiv 2024

  11. [11]

    S. Yuan, L. Shao, C. L. Yako, A. Gruebele, and J. K. Salisbury. Design and control of roller grasper v2 for in-hand manipulation. In2020 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), pages 9151–9158. IEEE, 2020

    2020

  12. [12]

    S. Yuan, S. Wang, R. Patel, M. Tippur, C. L. Yako, M. R. Cutkosky, E. Adelson, and K. Salisbury. Tactile-reactive roller grasper.IEEE Transactions on Robotics, 2025

    2025

  13. [13]

    G. Xie, R. Holladay, L. Chin, and D. Rus. In-hand manipulation with a simple belted parallel- jaw gripper.IEEE Robotics and Automation Letters, 9(2):1334–1341, 2024. doi:10.1109/LRA. 2023.3346750

    work page doi:10.1109/lra 2024

  14. [14]

    H. Wang, B. Gao, D. Zhao, and H. Shen. A reconfigurable gripper inspired by elastic belt for versatile in-hand manipulations.IEEE Robotics and Automation Letters, 10(9):9192–9199,

  15. [15]

    doi:10.1109/LRA.2025.3589879

    work page doi:10.1109/lra.2025.3589879 2025

  16. [16]

    X. Wang, L. Horrigan, J. Pinskier, G. Shi, V . Viswanathan, L. Liow, T. Bandyopadhyay, J. J. Chung, and D. Howard. Dexgrip: Multi-modal soft gripper with dexterous grasping and in-hand manipulation capacity. In2025 IEEE 8th International Conference on Soft Robotics (RoboSoft), pages 1–6. IEEE, 2025

    2025

  17. [17]

    He.GRIP-tape: A Novel Design of Tape-Based Manipulator

    G. He.GRIP-tape: A Novel Design of Tape-Based Manipulator. University of California, San Diego, 2023. 9

    2023

  18. [18]

    URL https://esun3dstoreeu.com/ products/tpu-95a

    esun etpu-95a 1.75mm 3d filament 1kg. URL https://esun3dstoreeu.com/ products/tpu-95a. 2025

    2025

  19. [19]

    Pinneri, S

    C. Pinneri, S. Sawant, S. Blaes, J. Achterhold, J. Stueckler, M. Rolinek, and G. Martius. Sample- efficient cross-entropy method for real-time planning. InConference on Robot Learning 2020,

    2020

  20. [20]

    URLhttps://corlconf.github.io/paper_217

  21. [21]

    Miolane, N

    N. Miolane, N. Guigui, A. L. Brigant, J. Mathe, B. Hou, Y . Thanwerdas, S. Heyder, O. Peltre, N. Koep, H. Zaatiti, H. Hajri, Y . Cabanes, T. Gerald, P. Chauchat, C. Shewmake, D. Brooks, B. Kainz, C. Donnat, S. Holmes, and X. Pennec. Geomstats: A python package for Riemannian geometry in machine learning.Journal of Machine Learning Research, 21(223):1–9, 2...

    2020

  22. [22]

    Y . Zhu, A. Joshi, P. Stone, and Y . Zhu. Viola: Imitation learning for vision-based manipulation with object proposal priors.arXiv preprint arXiv:2210.11339, 2022. doi:10.48550/arXiv.2210. 11339

    work page doi:10.48550/arxiv.2210 2022

  23. [23]

    E. Olson. AprilTag: A robust and flexible visual fiducial system. In2011 IEEE international conference on robotics and automation, pages 3400–3407. IEEE, 2011

    2011

  24. [24]

    Dex-Net 2.0: Deep Learning to Plan Robust Grasps with Synthetic Point Clouds and Analytic Grasp Metrics

    J. Mahler, J. Liang, S. Niyaz, M. Laskey, R. Doan, X. Liu, J. A. Ojea, and K. Goldberg. Dex-net 2.0: Deep learning to plan robust grasps with synthetic point clouds and analytic grasp metrics. arXiv preprint arXiv:1703.09312, 2017

    work page internal anchor Pith review Pith/arXiv arXiv 2017

  25. [25]

    O’Neill, A

    A. O’Neill, A. Rehman, A. Maddukuri, A. Gupta, A. Padalkar, A. Lee, A. Pooley, A. Gupta, A. Mandlekar, A. Jain, et al. Open x-embodiment: Robotic learning datasets and rt-x models: Open x-embodiment collaboration 0. In2024 IEEE International Conference on Robotics and Automation (ICRA), pages 6892–6903. IEEE, 2024

    2024

  26. [26]

    Black, N

    K. Black, N. Brown, J. Darpinian, K. Dhabalia, D. Driess, A. Esmail, M. R. Equi, C. Finn, N. Fusai, M. Y . Galliker, et al. \pi_{0.5}: a vision-language-action model with open-world generalization. In9th Annual Conference on Robot Learning, 2025

    2025

  27. [27]

    Bjorck, N

    NVIDIA, J. Bjorck, N. C. Fernando Castañeda, X. Da, R. Ding, L. J. Fan, Y . Fang, D. Fox, F. Hu, S. Huang, J. Jang, Z. Jiang, J. Kautz, K. Kundalia, L. Lao, Z. Li, Z. Lin, K. Lin, G. Liu, E. Llontop, L. Magne, A. Mandlekar, A. Narayan, S. Nasiriany, S. Reed, Y . L. Tan, G. Wang, Z. Wang, J. Wang, Q. Wang, J. Xiang, Y . Xie, Y . Xu, Z. Xu, S. Ye, Z. Yu, A....

    2025

  28. [28]

    T. Z. Zhao, V . Kumar, S. Levine, and C. Finn. Learning fine-grained bimanual manipulation with low-cost hardware.arXiv preprint arXiv:2304.13705, 2023

    work page internal anchor Pith review Pith/arXiv arXiv 2023

  29. [29]

    F. R. Hogan and A. Rodriguez. Reactive planar non-prehensile manipulation with hybrid model predictive control.The International Journal of Robotics Research, 39(7):755–773, 2020

    2020

  30. [30]

    Dawson-Haggerty et al. trimesh. URLhttps://trimesh.org/. 10 A Specification of the belt finger gripper To give an intuitive understanding of the in-hand manipulation capabilities of the Belt-Finger, the max- imum force and torques of each manipulation DoFs are measured with four types of objects on the test bench shown in Fig. 6. For each object, three di...

  31. [31]

    as the distance metric. It is also worth pointing out that although each target pose can be decomposed into translation and rotation in the gripper frame, this doesn’t guarantee a valid manipulation trajectory. Especially when the belt actions are not strictly mapping to the rotation or translation of the object due to the friction and the flexibility of ...