{"id":"ba589507-395e-46f0-93e6-d5c31085bc3d","arxiv_id":"2504.12967","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"The Krysalis Hand is a 790 g, 18-DoF five-finger robot hand with self-locking lead-screw joints that can passively hold over 10 lbs and actively push about 10 N per fingertip.","lead":"A team built a five-finger robot hand with 18 independently moving joints that weighs under 800 grams and can hold a 10 pound weight without using its motors. The key trick is a screw in each joint that locks itself, so holding force depends on the hand's strength, not motor power.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The '>10 lbs' payload claim rests entirely on an unreported FEA; the only direct test was a single capped 10 lb static hold, so the headline high-payload specification is not yet supported.","rationale":"The reader's weakest-assumption analysis correctly identifies the unreported FEA as the hinge for the '>10 lbs' payload claim. My own reading of the full text supports this: Section VI-A explicitly caps the physical test at 10 lbs and cites FEA only in a vague sentence, with no figures, tables, or equations presenting the analysis. The paper also contains a patent-pending note and a GitHub repository, but neither replaces the missing FEA evidence. The active force measurement of 'nearly 10 N' is at least directly measured, and the self-locking condition is analytically robust, so I do not see a reason to move the verdict to ACCEPT or REJECT. The existing CONDITIONAL verdict is appropriate: the hardware appears real and the mechanism is plausible, but the headline payload claim needs the missing FEA and an explicit overload test to be considered supported. My concern is the same as the reader's, hence agreement is 'agree' and no verdict adjustment is needed.","tokens_in":15531,"tokens_out":5907,"duration_ms":65470,"concrete_test":"Publish the FEA input deck (CAD geometry, Tough 1500 material card with measured as-printed modulus and strength, mesh, boundary conditions, and failure criterion) and perform an incremental static load-to-failure test on a spare assembled hand in the same configuration as Fig. 11, recording the failure load and failure location. If the measured failure load is below a pre-registered margin of 20% above 10 lbs, or if the failure occurs at a component or load level inconsistent with the FEA, revise Table I from '> 4.53 kg' to '>= 4.53 kg demonstrated' and mark the 'limited only by mechanical strength' claim as not validated.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The self-locking principle is the paper's central innovation, and the analytical check in Section III-B is sound: with a 0.35 mm lead on a 2.50 mm mean-diameter screw, the lead angle is about 2.55°, and even a conservative stainless/steel friction coefficient gives a friction angle far above that, so the screw should not back-drive. The load-bearing weakness is therefore not the self-locking calculation itself, but the extrapolation from 'held a 10 lb dumbbell once, deliberately capped' to the specification 'passive payload capacity > 4.53 kg' in Table I and the claim that payload is 'limited only by mechanical strength.' Section VI-A states that the load test was capped at 10 lbs to avoid damage, and that 'FEA results indicate' failure occurs under significantly higher loading. That FEA is not shown, not quantified, and not reproducible: there is no mesh, no material model for the printed Tough 1500 resin, no boundary conditions, no load case, no failure criterion, and no safety factor. Because most structural parts are SLA-printed, the difference between datasheet material properties and as-printed properties can be large, and an optimistic FEA could overstate the safety margin. If the true failure load is only slightly above 10 lbs, the 'high-payload' claim loses its distinguishing force relative to other lightweight hands. This is a correctness risk in the central claim, not a disagreement with community consensus.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper introduces the Krysalis Hand, a five-finger, 18-DoF anthropomorphic robotic end-effector with all actuators integrated into the fingers, palm, and a two-DoF wrist. The core design idea is a leadscrew-based finger actuation whose thread geometry is chosen to be self-locking, so that large external loads can be held passively without continuous motor torque. The reported specifications are a total mass of 790 g, a passive payload capacity greater than 10 lbs, an active fingertip force of about 10 N, and a finger speed of 91.5°/s. The manuscript describes the mechanical design, the self-locking analysis, the thumb CMC worm-gear mechanism, the coupled abduction-adduction drive, the wrist module, materials and electronics, and then presents experimental sections on fingertip force, thumb opposability, joint range of motion, grasping capability, and teleoperation via a MANUS glove and ROS 2. The central claimed contribution is that the passive payload limit is shifted from motor torque to structural strength, enabling lighter and cheaper motors while retaining high holding capacity.","tokens_in":15765,"tokens_out":3442,"duration_ms":38528,"significance":"If the specifications are substantiated, the design would be a valuable contribution to dexterous manipulation hardware: the self-locking leadscrew principle is clean, the hand is lightweight, and the proposed decoupling of passive payload from active motor force is genuinely useful for robot-learning applications. The analytical self-locking check in Section III-B is internally coherent, follows directly from the published geometry and a handbook friction coefficient, and does not use fitted constants; the release of control and teleoperation code is also a strength. However, the headline quantitative claims—especially the passive payload capacity greater than 10 lbs and the 10 N active force—currently rest on very limited experimental evidence, and the dexterity and teleoperation evaluations are qualitative. With additional experiments and reporting, the contribution could meet the bar for a strong hardware paper; in its present form the verification is too thin for the central specifications.","major_comments":[{"comment":"The specification 'Passive Payload Capacity > 4.53 kg (10 lbs)' is supported only by a single static hold of a 10 lb dumbbell, a test that the authors state was deliberately capped to avoid damage, plus an FEA that is mentioned but not reported. The FEA is not described in any reproducible detail: there is no mesh, no material model for the SLA Tough 1500 resin, no boundary conditions, no load case, no failure criterion, and no safety factor. Because the headline 'high-payload' claim depends on the assertion that structural failure occurs at 'significantly higher loading conditions,' this missing information is load-bearing. Please provide the full FEA setup and results, and support the specification with either repeated overload tests to failure or repeated capped tests with measured deflection and a conservative safety factor.","section":"Section VI-A and Table I"},{"comment":"The active fingertip force of 'nearly 10 N' is reported from a single time trace with no repeated trials, no error bars, and no description of the test protocol: the joint configuration, the fingertip contact point, the force sensor mounting, and whether the finger was stalled or moving at constant speed are not stated. Since Table I lists 10 N as a specification and the claim is used to argue that the hand decouples payload from active force, please report repeated trials with statistics and a precise protocol.","section":"Section VI-A and Fig. 10"},{"comment":"The dexterity and grasping evaluations are qualitative. Thumb opposition is reported as 'visually verified' contact success with no quantitative measure of accuracy or repeatability, and the grasping assessment consists of photographs categorized into precision, power, and tripod grasps without success rates, number of trials, or grasp-robustness metrics. Given that the introduction positions the hand for robotic learning and dexterous manipulation, quantitative task metrics (e.g., success rates over repeated trials, force during grasp, and object-feature coverage) are necessary to support these claims.","section":"Sections VI-B and VI-D"},{"comment":"The teleoperation section demonstrates a working pipeline but does not report any quantitative performance: there is no latency measurement, no tracking error between the MANUS glove and the hand joints, and no success rate for teleoperated manipulation. The text claims 'high-precision teleoperation,' which is not supported by the presented data. Please add either quantitative tracking/latency measurements or reframe the claim as a feasibility demonstration.","section":"Section VI-F"}],"minor_comments":[{"comment":"The abstract and Table I state that the passive payload is 'exceeding 10 lbs,' but the only direct test was capped at 10 lbs; please rephrase to 'at least 10 lbs' until overload data are provided.","section":"Abstract and Section VI-A"},{"comment":"The sentence 'Each phalange in the fingers D1 to D5 ... is driven by a micro DC motor with with embedded encoder' contains a duplicated 'with'; please correct.","section":"Section V"},{"comment":"The row formatting in Table II is inconsistent: for example, the 'Middle' finger block appears to have its joint labels shifted, and the 'Total RoM' under the Thumb block seems to belong to the Index block. Please recheck the alignment and labeling of all rows.","section":"Table II"},{"comment":"Reference [29] contains the placeholder '[insert page numbers]'; please complete the citation.","section":"References"},{"comment":"The manuscript says the GitHub repository contains control and teleoperation code, but it does not mention CAD or STL files for the mechanical parts. For a hardware paper, publishing the CAD files would substantially improve reproducibility; please state whether these files are available.","section":"Section IV and GitHub"},{"comment":"Equations (1) and (2) use a handbook friction coefficient for stainless steel rather than a measured value for the specific surface finish and nut material. The large margin (21.83° vs. 2.55°) makes the self-locking conclusion robust, but please state this assumption explicitly and briefly discuss the sensitivity to lubrication or wear.","section":"Section III-B"}],"recommendation":"major_revision","confidential_remarks":"The manuscript reads more like a strong conference hardware paper than a complete journal article. The central design idea and the self-locking analysis are sound, but the evaluation section needs substantially more rigor before the quantitative specifications can be accepted. The paper would benefit from either additional overload experiments and FEA reporting or a deliberate softening of the headline payload claim. I am not recommending rejection because the remaining issues are fixable within the manuscript's scope."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Here is my read on the Krysalis Hand paper. The genuinely new thing is per-phalanx leadscrew actuation with self-locking screws on all five fingers—no tendons, each phalanx independently driven. That is a real design departure from LEAP, Shadow, SCHUNK, and the HRI hand, and the prototype appears to work: 18 DoF, 790 g, a thumb CMC with a worm gear, and real demos including a power drill, handshake, and MANUS-glove teleop. The self-locking analysis is straightforward and correct: with a 0.35 mm lead on a 2.50 mm mean-diameter screw, the lead angle is 2.55° and the dry stainless-steel friction angle is about 21.8°, so the screw should not back-drive. That part holds up.\n\nThe paper also does some things well. The mechanical design is described in enough detail to reproduce the concept. They give the lead, diameter, materials, and joint ranges. They actually built the hand, and the GitHub repository with code is a plus. This is not a paper that only exists on paper.\n\nThe soft spot is the payload claim, and it is the headline number. The 'exceeding 10 lbs' in Table I and the 'limited only by mechanical strength' claim in Section VI-A rest on an FEA that is referenced but never shown—no mesh, no material model, no boundary conditions, no safety factor. The only direct measurement is a single static hold at 10 lbs, deliberately capped to avoid damage. That is honest, but it does not support the specification. The active force of 10 N is also a single trace with no repeats or error bars. The comparison to other hands is qualitative, not benchmarked. These are gaps in evidence, not proof the design is wrong.\n\nWho gets value from this? People building low-payload cobot hands, LfD researchers who need an affordable teleoperable end-effector, and anyone interested in self-locking transmission design. The reader's conditional verdict is right: the architecture is promising, but the specs need to be backed by released FEA, CAD, BOM, and repeated tests before the headline numbers are taken at face value.\n\nI would send this to peer review. It is a serious hardware contribution with a testable central claim. A good referee will ask for the missing FEA and repeat measurements, and the authors should be able to provide them. The core mechanism and working prototype deserve referee time.","headline":"A serious lightweight 18-DoF hand whose self-locking leadscrew design is sound and well-demonstrated, but whose headline '>10 lbs' payload spec currently rests on an unreported FEA and a single capped test.","tokens_in":16344,"tokens_out":2204,"would_cite":true,"duration_ms":23594,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"The paper claims a 790 gram, 18-DoF five-finger hand whose self-locking leadscrews let it passively hold over 10 lbs while each fingertip actively pushes about 10 N.","keywords":["robotic hand","dexterous manipulation","self-locking leadscrew","anthropomorphic end-effector","high payload","teleoperation","5-finger manipulator","18-DoF"],"falsifier":"Measure the load at which an unpowered, fully flexed Krysalis Hand joint first back-drives or fractures under a steadily increasing suspended weight, and compare it with the FEA-predicted failure load. If any joint creeps, back-drives, or breaks below 4.53 kg, the passive-payload claim is false; if the hand holds well beyond 10 lbs, the claim is supported.","tokens_in":15311,"feed_emoji":"🦾","tokens_out":6042,"duration_ms":62474,"temperature":0.7,"pith_summary":"This paper introduces the Krysalis Hand, a five-finger robotic hand that weighs 790 grams, has 18 actuated degrees of freedom, and is designed to hold over 10 lbs passively while each fingertip actively pushes about 10 N. The central claim is that self-locking leadscrew joints shift the payload limit from motor torque to the mechanical strength of the structure itself, so small, inexpensive motors can be used without sacrificing holding power. The authors report thumb opposability to all four fingers, successful precision, power, and tripod grasps, a total range of motion 11.2% greater than the average human hand across the measured joints, and real-time teleoperation with motion-capture gloves. A sympathetic reader would care because light weight, high DoF, high payload, and affordability are exactly the combination that has kept dexterous hands off low-payload collaborative arms and budget-constrained research labs.","feed_headline":"790-gram robotic hand lifts 10 lbs with motors off","feed_subtitle":"Self-locking leadscrews shift the payload limit from motor strength to structural strength, keeping a five-finger hand light and cheap.","key_machinery":"The load-bearing element is the self-locking leadscrew actuator repeated at every finger joint (DIP, PIP, and MCP) plus the worm-gear set at the thumb carpometacarpal joint. Self-locking is the condition $\\phi_s > \\alpha$: the friction angle of the screw-nut pair exceeds the lead angle of the screw thread, so an external load cannot rotate the screw backward. For the finger leadscrews the paper computes $\\alpha = 2.55^\\circ$ and $\\phi_s \\approx 21.83^\\circ$; for the thumb CMC worm gear it computes $\\alpha = 4.80^\\circ$ and $\\phi_s = 24.7^\\circ$. This mechanism carries the argument by converting holding torque from a motor burden into a structural one: the hand bears large passive loads with the motors off, while active fingertip force comes from the same screw driving a sliding nut that rotates the phalange through a two-link rocker.","core_discovery":"The central claim of the paper is that a fully actuated, non-tendon-driven five-finger hand can be both light and strong if each joint uses a self-locking leadscrew. Because the screw cannot be back-driven when the friction angle exceeds the lead angle, the hand holds a fixed pose without motor power; the passive payload limit then depends on the structural strength of the frame, not on the torque of its motors. The authors report a 790 g hand with 18 DoF that sustained a 10 lb static hold, nearly 10 N active force per fingertip, full thumb opposition to all four fingers, an average range of motion 11.2% greater than the human hand across the measured joints, and successful teleoperated grasping. The significance the authors draw is that payload capacity and actuation force are decoupled, so smaller and cheaper motors can be used while keeping high holding capacity.","pith_inferences":["The paper reports only a single capped 10 lb static test and refers to an unreported finite-element analysis for the true failure load; a natural next experiment is to load the unpowered hand to failure and confirm that FEA prediction, since the 'exceeding 10 lbs' claim rests entirely on that simulation.","The same self-locking screw-and-worm approach could plausibly be carried over to other energy-hungry joints, such as an elbow or waist, wherever sustained load holding matters, though dynamic loading and fatigue life would require separate validation.","If the passive/active decoupling holds up under repeated grasps and long-duration loads, the design could lower the cost of high-DoF hands enough to make dexterous manipulation practical for small-batch assembly and household robots, not only well-funded laboratories."],"forward_implications":["A hand built this way can hold a heavy object indefinitely without drawing motor current, which removes the overheating and power-consumption penalty of sustained grasps.","Because the motors only need to supply about 10 N per finger, the actuator set can be small and cheap enough to fit inside the 790 g hand, making the hand usable on low-payload collaborative arms.","With 18 DoF, thumb opposition to all fingertips, and two wrist axes, the hand covers precision, power, and tripod grasps over a range of everyday and industrial objects.","The demonstrated teleoperation pipeline offers a route for collecting human demonstrations in learning-from-demonstration systems without requiring a large, expensive, tendon-driven hand."],"supporting_citations":[{"why":"Supplies the engineering-statics self-locking condition that the friction angle must exceed the lead angle, which is the core mechanism behind the passive payload claim.","marker":"[51]"},{"why":"Supplies the dry stainless-steel friction coefficient used to compute the finger screw friction angle of about 21.83 degrees.","marker":"[52]"},{"why":"Supplies the 0.46 friction coefficient used for the 3D-printed aluminum worm/wheel pair in the thumb CMC self-locking calculation.","marker":"[54]"},{"why":"The cited open-source anthropomorphic hand used as the main comparison for size, weight, finger count, and DoF.","marker":"[32]"},{"why":"The heavyweight tendon-driven 20-DoF hand cited as the dominant weight and cost baseline that the Krysalis Hand aims to outperform.","marker":"[10]"},{"why":"The commercial five-finger hand with 20 joints but only 9 DoF, used as the baseline for dexterity limitations.","marker":"[37]"},{"why":"Provides the grasp taxonomy used to classify the precision, power, and tripod grasps in the evaluation.","marker":"[69]"},{"why":"Supplies the human-hand anthropometry data used to set the size and proportions of the Krysalis Hand.","marker":"[70]"},{"why":"Motion-capture glove hardware used in the teleoperation demonstration.","marker":"[73]"},{"why":"The teleoperation approach the paper's ROS 2 package was adapted from.","marker":"[71]"}],"fun_headline_variants":["Self-locking joints let 790g hand hold 10 lbs with motors off","Robotic hand: 790g, 18 DoF, holds 10 lbs without power","Self-locking screws shift hand payload from motors to structure","Dexterous 790g hand achieves 10lb hold via self-locking joints","Lightweight hand uses self-locking screws to decouple payload from motors"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The whole 'over 10 lbs' passive-payload figure rests on an unreported finite-element analysis; the only direct measurement is a single 10 lb static hold, deliberately capped to avoid breaking the hand.","fun_headline_variants_meta":{"raw":{"variants":["Self-locking joints let 790g hand hold 10 lbs with motors off","Robotic hand: 790g, 18 DoF, holds 10 lbs without power","Self-locking screws shift hand payload from motors to structure","Dexterous 790g hand achieves 10lb hold via self-locking joints","Lightweight hand uses self-locking screws to decouple payload from motors"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000724,"raw_usage":{"total_tokens":3263,"prompt_tokens":975,"completion_tokens":2288,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":591,"completion_tokens_details":{"reasoning_tokens":2184}},"tokens_in":591,"tokens_out":2288,"duration_ms":16301,"temperature":1.0,"reasoning_tokens":2184,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-16T12:18:32.497719+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Measure the load at which an unpowered, fully flexed Krysalis Hand joint first back-drives or fractures under a steadily increasing suspended weight, and compare it with the FEA-predicted failure load. If any joint creeps, back-drives, or breaks below 4.53 kg, the passive-payload claim is false; if the hand holds well beyond 10 lbs, the claim is supported.","supporting_citations":[{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Supplies the engineering-statics self-locking condition that the friction angle must exceed the lead angle, which is the core mechanism behind the passive payload claim."},{"cited_title":"Available: https://www.engineeringtoolbox.com/friction-coefficients-d 778.html","cited_arxiv_id":null,"evidence_quote":"Supplies the dry stainless-steel friction coefficient used to compute the finger screw friction angle of about 21.83 degrees."},{"cited_title":"Wang et al., ”Selective laser melting of aluminum and its alloys,” Materials, vol","cited_arxiv_id":null,"evidence_quote":"Supplies the 0.46 friction coefficient used for the 3D-printed aluminum worm/wheel pair in the thumb CMC self-locking calculation."},{"cited_title":"24, 2024","cited_arxiv_id":null,"evidence_quote":"The heavyweight tendon-driven 20-DoF hand cited as the dominant weight and cost baseline that the Krysalis Hand aims to outperform."},{"cited_title":"Available: https://schunk.com/de/de/greiftechnik/spezialgreifer/svh/c/PGR 3161","cited_arxiv_id":null,"evidence_quote":"The commercial five-finger hand with 20 joints but only 9 DoF, used as the baseline for dexterity limitations."},{"cited_title":"Feix et al., ”The grasp taxonomy of human grasp types,” IEEE Trans","cited_arxiv_id":null,"evidence_quote":"Provides the grasp taxonomy used to classify the precision, power, and tripod grasps in the evaluation."},{"cited_title":"Hand anthropometry of U.S. Army personnel,","cited_arxiv_id":null,"evidence_quote":"Supplies the human-hand anthropometry data used to set the size and proportions of the Krysalis Hand."},{"cited_title":"Available: https://www.manus-meta.com/products/quantum-metagloves","cited_arxiv_id":null,"evidence_quote":"Motion-capture glove hardware used in the teleoperation demonstration."}],"review_version":1}