pith. sign in

arxiv: 2605.15548 · v1 · pith:VCOKTAUGnew · submitted 2026-05-15 · 💻 cs.RO

KaRMA: A Kinematic Metric for Fine Manipulation Ability in Robotic Hands

Martin Peticco , Pulkit Agrawal This is my paper

Pith reviewed 2026-05-20 19:17 UTC · model grok-4.3

classification 💻 cs.RO
keywords robotic handsfine manipulationin-hand dexterityrolling contactkinematic metricreachability analysismanipulation abilitytwo-finger grasp
0
0 comments X

The pith

KaRMA measures reachable in-hand translations and rotations of a sphere through rolling motions to assess fine manipulation in robotic hands.

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

The paper introduces KaRMA to directly quantify dexterity as the ability to change an object's pose continuously while keeping contact, unlike traditional static metrics focused on workspace or grasp stability. KaRMA models this for a spherical object held in a two-finger precision pinch by searching reachable poses with a breadth-first exploration of small translation and rotation steps under rolling contact rules. It enforces joint limits, no collisions, and force closure at each step, then outputs three numbers: how much translation is possible, how much rotation, and how sensitive the result is to the starting grasp. Tests across sixteen common robotic hands show the scores split designs that static methods place in the same category and expose clear tradeoffs between translation and rotation that prior metrics miss. The results also line up with certain published task outcomes where Jacobian-style measures gave misleading signals.

Core claim

KaRMA is a kinematic-only metric that quantifies fine manipulation by finding all reachable in-hand poses of a spherical test object within a two-finger precision pinch through feasible rolling motions. It performs a breadth-first search over discrete translation and rotation primitives while enforcing joint limits, collision avoidance, rolling contact, and antipodal force feasibility, then reports translational coverage (KaRMA-T), rotational coverage (KaRMA-R), and grasp sensitivity (KaRMA-S). When applied to sixteen widely used robotic hands, the scores separate designs that rank identically under static proxies, surface translation-rotation tradeoffs invisible to existing baselines, and (

What carries the argument

Breadth-first search over translation and rotation primitives to map reachable poses of a spherical object under rolling contact constraints while respecting joint limits and force feasibility.

If this is right

  • Hands that appear equivalent under static metrics receive distinct KaRMA rankings based on their reachable rolling motions.
  • Clear translation-versus-rotation tradeoffs in dexterity appear that static workspace or manipulability measures hide.
  • KaRMA aligns better with certain published task benchmarks than Jacobian-based alternatives in cases where those alternatives mislead.
  • Hand designers gain a concrete kinematic target for improving continuous in-hand object control beyond initial grasp stability.

Where Pith is reading between the lines

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

  • Applying the same search approach to non-spherical objects or multi-contact grasps could test whether the current scores generalize to broader manipulation scenarios.
  • Incorporating KaRMA directly into automated hand design tools might produce new finger layouts optimized for rolling dexterity rather than only grasp force.
  • Combining KaRMA with existing dynamic or force metrics could create a fuller evaluation suite that covers both kinematic reach and contact stability.

Load-bearing premise

The reachable poses found for a spherical object under rolling contact in a two-finger precision pinch via breadth-first search adequately represent general fine manipulation ability across objects, grasps, and tasks.

What would settle it

A physical experiment showing that a hand with high KaRMA scores cannot perform fine in-hand reorientation tasks that a low-scoring hand can complete, or the reverse pattern.

Figures

Figures reproduced from arXiv: 2605.15548 by Martin Peticco, Pulkit Agrawal.

Figure 1
Figure 1. Figure 1: Overview of the KaRMA metric pipeline. Asterisk Test for two-finger translation [13], and planar ro￾tation benchmarks [14]. More broadly, other general manip￾ulation benchmarks involve performing a suite of tasks, often with learning-based methods, such as the Adroit dexterous manipulation suite [15]. These benchmarks are useful for evaluating a full system, but their outcomes depend heavily on controllers… view at source ↗
Figure 2
Figure 2. Figure 2: Schematic of the pinch grasp setup, including finger links, sphere, [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: KaRMA on two robot hands: Allegro and Sharpa. Red denotes [PITH_FULL_IMAGE:figures/full_fig_p005_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: KaRMA-T (circles) and KaRMA-R (squares) vs workspace [PITH_FULL_IMAGE:figures/full_fig_p006_4.png] view at source ↗
read the original abstract

Traditional robotic hand metrics focus on static properties such as workspace, manipulability, and grasp stability. However, these metrics do not directly measure dexterity under the standard definition in robotic manipulation: the ability to continuously change an object's pose within the hand while maintaining contact from an initial grasp. We introduce Kinematic Rolling Manipulation Ability (KaRMA), a kinematic-only metric for fine manipulation that quantifies reachable in-hand translation and reorientation of a spherical test object within a two-finger precision pinch through feasible rolling motions. KaRMA enforces joint limits, collision constraints, rolling contact, and antipodal force feasibility, then investigates reachable in-hand object poses via breadth-first search over translation and rotation primitives. KaRMA reports three scores: translational coverage (KaRMA-T), rotational coverage (KaRMA-R), and sensitivity to the initial grasp (KaRMA-S). We evaluate KaRMA on 16 widely used robotic hands and compare against static baselines, showing that KaRMA separates hands that rank identically under static proxies, reveals translation-rotation tradeoffs invisible to existing baselines, and is qualitatively consistent with selected published task benchmarks where Jacobian-based metrics can be misleading.

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 / 2 minor

Summary. The manuscript proposes Kinematic Rolling Manipulation Ability (KaRMA) as a kinematic metric for fine manipulation dexterity in robotic hands. It defines reachable in-hand translations and reorientations of a spherical test object in a two-finger precision pinch under rolling contact, enforcing joint limits, collisions, rolling constraints, and antipodal force feasibility. Reachable poses are explored via breadth-first search over discrete translation and rotation primitives. Three scores are computed: translational coverage (KaRMA-T), rotational coverage (KaRMA-R), and sensitivity to initial grasp (KaRMA-S). Evaluated on 16 robotic hands, the paper claims KaRMA separates hands ranked identically by static proxies, reveals translation-rotation tradeoffs invisible to baselines, and shows qualitative consistency with selected task benchmarks where Jacobian metrics can mislead.

Significance. If the modeling assumptions hold, KaRMA would supply a direct, non-circular kinematic measure of continuous in-hand dexterity that complements existing static metrics such as workspace volume or manipulability ellipsoids. A clear strength is the parameter-free construction: the metric is obtained from explicit constraint satisfaction and exhaustive search over primitives rather than fitted parameters or self-referential quantities, supporting reproducibility. This approach could help designers distinguish hands for tasks requiring sustained rolling-based reorientation and could highlight tradeoffs not captured by purely static or Jacobian-based proxies.

major comments (2)
  1. [§3] §3 (KaRMA definition and reachable-set computation): The metric is constructed exclusively from reachable poses of a spherical object under rolling contact in an antipodal two-finger precision pinch, obtained via BFS over translation/rotation primitives. This specific proxy is load-bearing for the central claims of separation, tradeoff revelation, and benchmark consistency; the manuscript does not test whether hand rankings or tradeoff structure remain stable for non-spherical objects, sliding contacts, multi-finger coordination, or different initial grasps. A direct check would be to recompute the three KaRMA scores for at least one alternative object or contact model and report whether orderings change.
  2. [§5] §5 (Evaluation and comparisons): The reported qualitative consistency with selected published task benchmarks is presented without quantitative agreement metrics, error analysis, statistical tests, or explicit criteria for benchmark selection. This leaves the claim that KaRMA aligns where Jacobian-based metrics mislead without numerical support, weakening the evidential basis for the separation and tradeoff assertions.
minor comments (2)
  1. [§3] The three scores KaRMA-T, KaRMA-R, and KaRMA-S are introduced clearly in the abstract and §3, but a compact summary table listing their exact definitions, normalization, and coverage interpretation would improve readability.
  2. [Figures] Figure captions and axis labels in the reachable-pose visualizations could more explicitly distinguish translational versus rotational coverage components to aid interpretation of the reported tradeoffs.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the constructive feedback on our manuscript. We have carefully considered the major comments and provide point-by-point responses below. Where appropriate, we propose revisions to address the concerns raised.

read point-by-point responses

  1. Referee: [§3] §3 (KaRMA definition and reachable-set computation): The metric is constructed exclusively from reachable poses of a spherical object under rolling contact in an antipodal two-finger precision pinch, obtained via BFS over translation/rotation primitives. This specific proxy is load-bearing for the central claims of separation, tradeoff revelation, and benchmark consistency; the manuscript does not test whether hand rankings or tradeoff structure remain stable for non-spherical objects, sliding contacts, multi-finger coordination, or different initial grasps. A direct check would be to recompute the three KaRMA scores for at least one alternative object or contact model and report whether orderings change.

    Authors: We appreciate the referee's point regarding the specificity of our proxy. The choice of a spherical object under rolling contact in a two-finger precision pinch is deliberate, as it isolates the kinematic ability for continuous pose change via rolling without introducing object-specific geometric features or requiring sliding. This provides a standardized, parameter-free measure focused on fine manipulation dexterity. While we acknowledge that extending to non-spherical objects or multi-finger setups would be valuable for broader validation, such extensions would require substantial additional modeling and computation. We will revise the manuscript to include a dedicated discussion on the assumptions and limitations of the current proxy, along with plans for future extensions to alternative contact models. revision: partial

  2. Referee: [§5] §5 (Evaluation and comparisons): The reported qualitative consistency with selected published task benchmarks is presented without quantitative agreement metrics, error analysis, statistical tests, or explicit criteria for benchmark selection. This leaves the claim that KaRMA aligns where Jacobian-based metrics mislead without numerical support, weakening the evidential basis for the separation and tradeoff assertions.

    Authors: We agree that providing more quantitative support would strengthen the presentation. However, the selected task benchmarks come from heterogeneous experimental setups in the literature, making direct quantitative agreement metrics difficult without re-implementing the tasks on our hands. We will revise §5 to include explicit criteria for benchmark selection, additional details on the qualitative comparisons, and a discussion of why statistical tests are not applicable here. We believe this will better support the claims without overclaiming numerical agreement. revision: partial

Circularity Check

0 steps flagged

KaRMA derivation is self-contained kinematic computation with no reduction to inputs or self-citations.

full rationale

The paper explicitly constructs KaRMA as the set of reachable in-hand poses for a spherical object under rolling contact in a two-finger precision pinch, obtained by breadth-first search over translation and rotation primitives while enforcing joint limits, collisions, rolling constraints, and antipodal force feasibility. The reported scores KaRMA-T, KaRMA-R, and KaRMA-S are direct outputs of this search procedure rather than fitted parameters or quantities defined in terms of the final metric. No load-bearing steps invoke self-citations, uniqueness theorems from prior author work, or ansatzes smuggled via citation; the comparisons to static baselines and selected task benchmarks are external evaluations of the computed values. The modeling choice of sphere, rolling-only contact, and two-finger pinch is a deliberate scope limitation rather than a circular premise that forces the reported separations or tradeoffs.

Axiom & Free-Parameter Ledger

0 free parameters · 2 axioms · 0 invented entities

The metric rests on domain assumptions about rolling contact and antipodal feasibility plus the modeling choice of a spherical test object; no free parameters or invented physical entities are described in the abstract.

axioms (2)
  • domain assumption Rolling contact without slip is the dominant feasible motion for continuous in-hand pose change in a precision pinch grasp.
    KaRMA is constructed exclusively around rolling primitives and enforces rolling contact as a hard constraint.
  • domain assumption Breadth-first search over discrete translation and rotation primitives sufficiently explores the reachable set under the stated constraints.
    The reachable-pose investigation step relies on this search procedure.

pith-pipeline@v0.9.0 · 5735 in / 1398 out tokens · 47232 ms · 2026-05-20T19:17:26.031387+00:00 · methodology

discussion (0)

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

Lean theorems connected to this paper

Citations machine-checked in the Pith Canon. Every link opens the source theorem in the public Lean library.

  • IndisputableMonolith/Foundation/RealityFromDistinction.lean reality_from_one_distinction unclear
    ?
    unclear

    Relation between the paper passage and the cited Recognition theorem.

    KaRMA measures the reachable set of object translations and orientations from an initial two-finger precision pinch on a spherical test object under rolling contact... reports three scores: translational ability (KaRMA-T), rotational ability (KaRMA-R), and initial-grasp sensitivity (KaRMA-S).

What do these tags mean?
matches
The paper's claim is directly supported by a theorem in the formal canon.
supports
The theorem supports part of the paper's argument, but the paper may add assumptions or extra steps.
extends
The paper goes beyond the formal theorem; the theorem is a base layer rather than the whole result.
uses
The paper appears to rely on the theorem as machinery.
contradicts
The paper's claim conflicts with a theorem or certificate in the canon.
unclear
Pith found a possible connection, but the passage is too broad, indirect, or ambiguous to say the theorem truly supports the claim.

Reference graph

Works this paper leans on

27 extracted references · 27 canonical work pages

  1. [1]

    Hands for dexterous manipulation and robust grasping: a difficult road toward simplicity,

    A. Bicchi, “Hands for dexterous manipulation and robust grasping: a difficult road toward simplicity,”IEEE Transactions on Robotics and Automation, vol. 16, no. 6, pp. 652–662, 2000

    work page 2000

  2. [2]

    Dexterity measures for the design and control of kinematically redundant manipulators,

    C. A. Klein and B. E. Blaho, “Dexterity measures for the design and control of kinematically redundant manipulators,”The International Journal of Robotics Research, vol. 6, no. 2, pp. 72–83, 1987. [Online]. Available: https://doi.org/10.1177/027836498700600206

    work page doi:10.1177/027836498700600206 1987

  3. [3]

    Dexterous grippers: Putting nonholonomy to work for fine manipulation,

    A. Bicchi and A. Marigo, “Dexterous grippers: Putting nonholonomy to work for fine manipulation,”The International Journal of Robotics Research, vol. 21, no. 5-6, pp. 427–442, 2002. [Online]. Available: https://doi.org/10.1177/027836402321261968

    work page doi:10.1177/027836402321261968 2002

  4. [4]

    The role of the thumb: Study of finger motion in grasping and reachability space in human and robotic hands,

    G. Cotugno, K. Althoefer, and T. Nanayakkara, “The role of the thumb: Study of finger motion in grasping and reachability space in human and robotic hands,” in2014 IEEE International Conference on Robotics and Automation (ICRA), 2014, pp. 5345–5351

    work page 2014

  5. [5]

    Manipulability of robotic mechanisms,

    T. Yoshikawa, “Manipulability of robotic mechanisms,”The Interna- tional Journal of Robotics Research, vol. 4, no. 2, pp. 3–9, 1985

    work page 1985

  6. [6]

    Kinematic design optimization for anthropomorphic robot hand based on interactivity of fingers,

    W. S. You, Y . H. Lee, G. Kang, H. S. Oh, J. K. Seo, and H. R. Choi, “Kinematic design optimization for anthropomorphic robot hand based on interactivity of fingers,”Intell. Serv. Robot., vol. 12, no. 2, p. 197–208, Apr. 2019. [Online]. Available: https://doi.org/10.1007/s11370-019-00274-x

    work page doi:10.1007/s11370-019-00274-x 2019

  7. [7]

    On dexterity and dexterous manipula- tion,

    R. R. Ma and A. M. Dollar, “On dexterity and dexterous manipula- tion,” in2011 15th International Conference on Advanced Robotics (ICAR), 2011, pp. 1–7

    work page 2011

  8. [8]

    Workshop on the design and control of dextrous hands,

    J. M. Hollerbach, “Workshop on the design and control of dextrous hands,” Massachusetts Institute of Technology, Artificial Intelligence Laboratory, Cambridge, MA, Tech. Rep. AIM-661, April 1982. [Online]. Available: http://hdl.handle.net/1721.1/5688

    work page 1982

  9. [9]

    Clinical test of apposition and counter-apposition of the thumb,

    A. Kapandji, “Clinical test of apposition and counter-apposition of the thumb,”Annales de Chirurgie de la Main, vol. 5, no. 1, pp. 67–73, 1986

    work page 1986

  10. [10]

    Analysis of multifingered hands,

    J. Kerr and B. Roth, “Analysis of multifingered hands,”The International Journal of Robotics Research, vol. 4, no. 4, pp. 3–17, 1986. [Online]. Available: https://doi.org/10.1177/ 027836498600400401

    work page 1986

  11. [11]

    Anthropo- morphic robot hand: Gifu hand iii,

    T. Mouri, H. Kawasaki, K. Yoshikawa, J. Takai, and S. Ito, “Anthropo- morphic robot hand: Gifu hand iii,” inProceedings of the International Conference on Control, Automation and Systems (ICCAS), 2002, pp. 1288–1293

    work page 2002

  12. [12]

    The elliott and connolly benchmark: A test for evaluating the in-hand dexterity of robot hands,

    R. Coulson, C. Li, C. Majidi, and N. S. Pollard, “The elliott and connolly benchmark: A test for evaluating the in-hand dexterity of robot hands,” in2020 IEEE-RAS 20th International Conference on Humanoid Robots (Humanoids), 2021, pp. 238–245

    work page 2021

  13. [13]

    Bench- marking a robot hand’s ability to translate objects using two fingers,

    J. Morrow, J. Campbell, R. Balasubramanian, and C. Grimm, “Bench- marking a robot hand’s ability to translate objects using two fingers,” IEEE Robotics and Automation Letters, vol. 7, no. 1, pp. 588–593, 2022

    work page 2022

  14. [14]

    Benchmarking planar rotation capabilities of robot hands with fingers,

    ——, “Benchmarking planar rotation capabilities of robot hands with fingers,” in2023 20th International Conference on Ubiquitous Robots (UR), 2023

    work page 2023

  15. [15]

    Learning complex dexterous manipulation with deep reinforcement learning and demonstrations,

    A. Rajeswaran, V . Kumar, A. Gupta, G. Vezzani, J. Schulman, E. Todorov, and S. Levine, “Learning complex dexterous manipulation with deep reinforcement learning and demonstrations,” inProceedings of Robotics: Science and Systems (RSS), 2018

    work page 2018

  16. [16]

    The kinematics of contact and grasp,

    D. J. Montana, “The kinematics of contact and grasp,”The Interna- tional Journal of Robotics Research, vol. 7, no. 3, pp. 17–32, 1988

    work page 1988

  17. [17]

    Dexterous manipulation through rolling,

    A. Bicchi and R. Sorrentino, “Dexterous manipulation through rolling,” inProceedings of the IEEE International Conference on Robotics and Automation (ICRA), 1995, pp. 452–457

    work page 1995

  18. [18]

    Relaxed-rigidity constraints: In- grasp manipulation using purely kinematic trajectory optimization,

    B. Sundaralingam and T. Hermans, “Relaxed-rigidity constraints: In- grasp manipulation using purely kinematic trajectory optimization,” in Proceedings of Robotics: Science and Systems (RSS), 2017

    work page 2017

  19. [19]

    Dexterous manipula- tion graphs,

    S. Cruciani, C. Smith, D. Kragic, and K. Hang, “Dexterous manipula- tion graphs,” in2018 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), 2018, pp. 2040–2047

    work page 2018

  20. [20]

    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 Transactions on Human-Machine Systems, vol. 46, no. 1, pp. 66–77, 2016

    work page 2016

  21. [21]

    Dimensional synthesis of three-fingered robot hands for maximal precision manipulation workspace,

    J. Borr `as and A. M. Dollar, “Dimensional synthesis of three-fingered robot hands for maximal precision manipulation workspace,”The International Journal of Robotics Research, vol. 34, no. 14, pp. 1731–1746, Dec. 2015. [Online]. Available: https://journals.sagepub. com/doi/10.1177/0278364915608250

    work page doi:10.1177/0278364915608250 2015

  22. [22]

    HEALPix: A framework for high- resolution discretization and fast analysis of data distributed on the sphere,

    K. M. G ´orski, E. Hivon, A. J. Banday, B. D. Wandelt, F. K. Hansen, M. Reinecke, and M. Bartelmann, “HEALPix: A framework for high- resolution discretization and fast analysis of data distributed on the sphere,”The Astrophysical Journal, vol. 622, no. 2, pp. 759–771, 2005

    work page 2005

  23. [23]

    OSQP: An operator splitting solver for quadratic programs,

    B. Stellato, G. Banjac, P. Goulart, A. Bemporad, and S. Boyd, “OSQP: An operator splitting solver for quadratic programs,”Mathematical Programming Computation, vol. 12, no. 4, pp. 637–672, 2020

    work page 2020

  24. [24]

    A musculoskeletal model of the hand and wrist capable of simulating functional tasks,

    D. C. McFarland, B. I. Binder-Markey, J. A. Nichols, S. J. Wohlman, M. de Bruin, and W. M. Murray, “A musculoskeletal model of the hand and wrist capable of simulating functional tasks,”IEEE Transactions on Biomedical Engineering, vol. 70, no. 5, pp. 1599–1610, 2023

    work page 2023

  25. [25]

    Mri- based skeletal hand movement model,

    G. Stillfried, U. Hillenbrand, M. Settles, and P. van der Smagt, “Mri- based skeletal hand movement model,” inThe Human Hand - A Source of Inspiration for Robotic Hands, ser. Springer Tracts on Advanced Robotics. Springer, 2013, vol. 95, pp. 59–75

    work page 2013

  26. [26]

    DexMachina: Functional retargeting for bimanual dexterous manipulation,

    M. Zhao, Y . Hou, D. Fox, Y . Narang, A. Mandlekar, and S. Song, “DexMachina: Functional retargeting for bimanual dexterous manipulation,” 2025. [Online]. Available: https://arxiv.org/abs/2505. 24853

    work page 2025

  27. [27]

    Isyhand: A dexterous multi-finger robot hand with an articulated palm,

    B. A. Richardson, F. Gr ¨uninger, L. Mack, J. Stueckler, and K. J. Kuchenbecker, “Isyhand: A dexterous multi-finger robot hand with an articulated palm,” inIEEE-RAS International Conference on Hu- manoid Robots (Humanoids), Seoul, Korea, Sep. 2025

    work page 2025