{"id":"9df15983-aead-441d-83f9-d02924e5d12f","arxiv_id":"2605.15548","paper_version":1,"verdict":"UNVERDICTED","confidence":"LOW","novelty_score":7.0,"correctness_risk":"unknown","formal_verification":"none","parameter_count":0,"one_line_summary":"KaRMA is a kinematic metric that quantifies reachable in-hand translation and reorientation of a spherical object in robotic hands via constrained rolling motions and reports translational coverage, rotational coverage, and sensitivity to initial grasp.","lead":"The paper introduces KaRMA, a new kinematic metric that scores robotic hands on how far they can translate and rotate a spherical object inside a two-finger pinch grasp using only rolling motions while respecting joint limits and contacts. A smart generalist might read it to better understand or select hands for tasks that require in-hand dexterity rather than just static grasping ability.","discovery_kind":"new_method","skeptic_critique":{"model":"grok-4.3","headline":"KaRMA's validity as a general fine-manipulation metric rests on whether reachable poses for a sphere under rolling in a two-finger pinch (via BFS over primitives) proxy dexterity across objects, grasps, and tasks.","rationale":"The reader's weakest assumption directly identifies the load-bearing premise in the metric definition. No internal inconsistency or missing formal verification is apparent from the given description, but the broad claims require this specific proxy to be representative. The proposed test would falsify or support that premise with minimal additional computation.","tokens_in":1744,"tokens_out":355,"duration_ms":34574,"concrete_test":"Recompute KaRMA scores for the same 16 hands but replace the sphere with a cylinder (allowing limited sliding at contact points) and rerun the BFS search; if hand orderings or KaRMA-T vs. KaRMA-R tradeoffs shift by more than one position for at least three hands, the generalization assumption does not hold.","verdict_should_be":"CONDITIONAL","load_bearing_attack":"The central claim—that KaRMA separates hands identically ranked by static proxies, exposes translation-rotation tradeoffs, and aligns with task benchmarks—requires that the specific construction (spherical test object, rolling contact only, antipodal force check, two-finger precision pinch, and discrete BFS over translation/rotation primitives) adequately captures the continuous reachable set for general in-hand manipulation. This premise is baked into the definitions of KaRMA-T, KaRMA-R, and KaRMA-S. If reachable sets for non-spherical objects, sliding contacts, multi-finger coordination, or different initial grasps differ in ranking or tradeoff structure, the reported separations and consistencies would not generalize.","agreement_with_reader":"agree"},"referee_report":{"model":"grok-4.3","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.","tokens_in":1911,"tokens_out":696,"duration_ms":48868,"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":[{"comment":"§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.","section":"§3"},{"comment":"§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.","section":"§5"}],"minor_comments":[{"comment":"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.","section":"§3"},{"comment":"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.","section":"Figures"}],"recommendation":"major_revision","confidential_remarks":"The manuscript fits the scope of a robotics journal focused on manipulation and control. The limited validation scope (single object class and grasp type) is the primary concern that revisions should address; once supplemented, the work would be a stronger candidate for publication."},"author_rebuttal":{"model":"grok-4.3","summary":"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.","responses":[{"response":"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_made":"partial","referee_comment":"[§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."},{"response":"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_made":"partial","referee_comment":"[§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."}],"tokens_in":1526,"tokens_out":532,"duration_ms":42287,"standing_objections":[]},"desk_editor":{"model":"grok-4.3","letter":"KaRMA introduces a kinematic metric that quantifies how far a robotic hand can translate and reorient a sphere through rolling motions from an initial two-finger pinch. The search uses breadth-first exploration of translation and rotation primitives while checking joint limits, collisions, rolling contact, and antipodal forces. This produces three scores that separate hands ranked the same by workspace or Jacobian metrics and point out translation-rotation tradeoffs those baselines miss.","headline":"KaRMA gives a search-based kinematic score for in-hand sphere rolling that splits some hands static metrics treat as equal and flags translation-rotation tradeoffs, but the whole thing is tied to one narrow test case.","tokens_in":2376,"tokens_out":175,"would_cite":false,"duration_ms":44835,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":{"model":"grok-4.3","evidence":[{"relation":"unclear","rs_module":"IndisputableMonolith/Foundation/RealityFromDistinction.lean","rs_theorem":"reality_from_one_distinction","paper_passage":"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)."}],"headline":"KaRMA kinematic reachability metric for robotic hands lies outside RS scope","alignment":"orthogonal","rationale":"The paper's core construction (BFS over translation/rotation primitives with rolling-contact QP, joint/collision/antipodal constraints, HEALPix orientation bins, Lref scaling, and KaRMA-T/R/S scores) is a practical engineering metric for two-finger pinch dexterity. It contains none of the RS-shaped structures (J-cost, φ-ladder, 8-tick periodicity, ratio-symmetric forcing, or parameter-free constant derivations) and makes no claims that intersect the reality_from_one_distinction chain or its downstream theorems.","tokens_in":50986,"confidence":"high","tokens_out":269,"duration_ms":12873,"cache_read_input_tokens":38528,"cache_creation_input_tokens":0},"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"grok-4.3","headline":"KaRMA measures reachable in-hand translations and rotations of a sphere through rolling motions to assess fine manipulation in robotic hands.","keywords":["robotic hands","fine manipulation","in-hand dexterity","rolling contact","kinematic metric","reachability analysis","manipulation ability","two-finger grasp"],"falsifier":"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.","tokens_in":2633,"feed_emoji":"🤖","tokens_out":776,"duration_ms":38977,"temperature":0.7,"pith_summary":"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.","feed_headline":"KaRMA ranks robot hands by reachable rolling motions","feed_subtitle":"The metric finds how far a sphere can translate and rotate in a two-finger pinch, splitting hands that static tests rate the same.","key_machinery":"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.","core_discovery":"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 (","pith_inferences":["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."],"forward_implications":["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."],"fun_headline_variants":["KaRMA measures rolling reach for robot hands","Hands ranked by in-hand sphere rolling ability","KaRMA separates robot hands on rolling coverage","Metric reveals tradeoffs in manipulation dexterity"],"cache_read_input_tokens":64,"weakest_assumption_plain":"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.","fun_headline_variants_meta":{"raw":{"variants":["KaRMA measures rolling reach for robot hands","Hands ranked by in-hand sphere rolling ability","KaRMA separates robot hands on rolling coverage","Metric reveals tradeoffs in manipulation dexterity"]},"model":"grok-4.3","cost_usd":0.011165,"raw_usage":{"total_tokens":4849,"prompt_tokens":712,"num_sources_used":0,"completion_tokens":54,"cost_in_usd_ticks":111653000,"prompt_tokens_details":{"text_tokens":712,"audio_tokens":0,"image_tokens":0,"cached_tokens":64},"completion_tokens_details":{"audio_tokens":0,"reasoning_tokens":4083,"accepted_prediction_tokens":0,"rejected_prediction_tokens":0}},"tokens_in":712,"tokens_out":54,"duration_ms":48756,"temperature":1.0,"reasoning_tokens":4083,"cache_read_input_tokens":64,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-05-20T19:17:26.031387+00:00","model_set":{"reader":"grok-4.3"},"falsifier":"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.","supporting_citations":[],"review_version":1}