{"id":"e2bf7cda-0b66-480c-b641-421631d2aac9","arxiv_id":"2509.10862","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":5.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":3,"one_line_summary":"Using pulley efficiency data from the new testing machine in a tension allocation controller reduced a wire-driven robot's end-effector force root mean square error from 10.9 N to 9.5 N, about 13%.","lead":"This paper presents a wire testing machine that pre-stretches wires, measures how much tension is lost over pulleys of different diameters, and records dynamic behavior when wire length changes, then applies that data to reduce force error in a wire-driven robot. Wire-driven robots are light and safe but hard to control because wires stretch and lose tension, so a calibration and compensation tool could make them more practical.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The 13% force-error reduction is supported by five trials per condition with no variance or significance test; if per-trial RMSEs show overlap, the central claim is unsupported.","rationale":"The paper's central claim is that testing-machine data improves force control, evidenced by the 13% RMSE reduction in Section IV-C. The most load-bearing condition for that claim is that the observed improvement is real, not noise. Since no variance information is given, the reader cannot assess this. Unlike the transferability issue, which the authors explicitly acknowledge in the conclusion as future work, the missing statistical analysis is not addressed anywhere. The transferability concern would matter for a claim of universality, but the empirical comparison in Section IV-C is self-contained: it directly compares two controllers on the same robot. Therefore, the statistical validity is the gate. If the statistical check passes, the experiment supports the claim on this robot, and transferability can be handled as a stated limitation. If it fails, the central claim collapses. The reader already conditioned the verdict on releasing data with error bars, so my concern aligns with the reader's rationale, though the reader's named weakest assumption is transferability and mine is statistical robustness.","tokens_in":8347,"tokens_out":9005,"duration_ms":81561,"concrete_test":"Request the per-trial RMSE values (or full time-series) for the five with-compensation and five without-compensation trials from Section IV-C, and perform a paired t-test or Wilcoxon signed-rank test on the per-trial RMSE differences. Report the 95% confidence interval for the mean difference. If the 1.4 N reduction is not significant at alpha=0.05, the claim that the wire testing machine data improves force control accuracy is not supported by the experiment.","verdict_should_be":"UNCHANGED","load_bearing_attack":"Section IV-C reports that tension-loss compensation reduced vertical end-effector force RMSE from 10.9 N to 9.5 N, a 13% reduction, based on five trials per condition (Fig. 11). No standard deviations, confidence intervals, or significance tests are reported, so the difference of 1.4 N could be within trial-to-trial variability. The figure's two curves appear to overlap substantially over time, so the RMSE difference may be dominated by a particular phase of the ramp rather than a consistent reduction. Since the paper's central quantitative claim rests entirely on these two mean RMSE values, the missing statistical support is the most load-bearing weakness. A secondary concern, acknowledged by the authors in Section V, is that the per-pulley efficiency was measured at a fixed 90 degree wrap angle and only at 200 N and 400 N, while the robot's pulleys may see other wrap angles and tensions; but this affects generality, not the internal validity of the comparison.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper proposes a Universal Wire Testing Machine with three functions: pre-stretching wires to remove initial plastic elongation, measuring tension transmission efficiency for combinations of four wire types and eight pulley diameters, and measuring tension dynamics under variable total wire length. The authors report efficiency trends, characterize variable-length versus constant-length frequency responses, and apply the measured per-pulley efficiency in a quadratic-program tension allocator (Eq. (7)) for force control of a wire-driven robot. The central quantitative claim is that this compensation reduced the vertical end-effector force RMSE from 10.9 N to 9.5 N, about 13%, based on five trials per condition.","tokens_in":8521,"tokens_out":3719,"duration_ms":36184,"significance":"The hardware contribution is useful: the testing machine covers a wider range of pulley diameters and wire types than prior studies, and the efficiency trends—increasing with pulley diameter and decreasing with wire diameter—are physically sensible and consistent with the reported measurement curves. The dynamic variable-length wire measurements also address a real gap in the literature. However, the paper's main claim that the testing-machine data improve robot force control rests on two mean RMSE values with no reported uncertainty, and the transfer of a single per-pulley efficiency to all robot operating conditions is not established. If these issues are resolved, the work could be a solid practical contribution to wire-driven robot design and control; as it stands, its significance is contingent.","major_comments":[{"comment":"The claimed 13% force-error reduction is supported only by the difference between two mean RMSE values, 10.9 N without compensation and 9.5 N with compensation, with five trials per condition and no standard deviation, confidence interval, or significance test. The two curves in Fig. 11 appear to overlap substantially over the trial, so the mean difference could be dominated by a short phase of the ramp rather than a consistent reduction. Please report per-trial RMSE values with dispersion and a statistical comparison, or explicitly soften the claim to a preliminary observation.","section":"Section IV-C, Fig. 11"},{"comment":"The compensation model uses a single measured per-pulley efficiency, obtained at a fixed 90-degree wrap angle and at 200 N and 400 N input tensions, as an exponentiated efficiency matrix for every passive pulley in the robot, independent of actual tension, wrap angle, and direction of motion. Section V acknowledges the wrap-angle limitation, but the robot wires may also reverse direction, and prior work [16] shows that direction changes introduce friction not captured by constant-efficiency models. The authors should either demonstrate that the test-bench conditions are representative of the robot's operating conditions or restrict the claim accordingly; otherwise the 9.5 N versus 10.9 N comparison may not transfer beyond the specific configuration tested.","section":"Section IV-A and Eq. (5)"},{"comment":"The experimental description does not state the robot posture, the range of wire tensions actually generated, or whether the five trials in each condition used identical trajectories and initial states. Without this information, it is difficult to assess whether the reported RMSE difference is attributable to the compensation term rather than to trial-specific variability or configuration differences.","section":"Section IV-C"}],"minor_comments":[{"comment":"The text reports a plastic deformation of 0.43 m, while the Fig. 4 caption states that the wire elongated from 8.2 m to 8.6 m, which is 0.4 m; please reconcile these numbers.","section":"Section III-B, Fig. 4"},{"comment":"Twenty measurements per condition were averaged to produce each efficiency point, but no error bars or dispersion measures are shown; providing them would help assess the reliability of the reported trends.","section":"Section IV-A, Figs. 7 and 8"},{"comment":"The reference list is not in first-citation order (e.g., [4] appears before [2] and [3]), and the in-text citation of [16] as 'Sung-Hyun et al.' should be checked against the actual author names; standard numeric citation practice would also avoid this ambiguity.","section":"References"},{"comment":"The frequency-response comparison would be clearer if the authors defined exactly what distinguishes the 'variable-length' from the 'constant-length' condition beyond the linear loading unit being free or fixed, and if they reported whether the plotted curves are from single trials or averaged over repetitions.","section":"Section IV-B, Fig. 9"}],"recommendation":"major_revision","confidential_remarks":"The manuscript is within the scope of a robotics venue and the hardware contribution is genuine. The main reason for major revision rather than rejection is that the central quantitative claim is potentially supportable but currently lacks statistical grounding and transferability evidence; both are fixable with additional analysis and experiments."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Colleague, this paper is worth your time for the hardware and the dataset, not for the headline robot demo. The testing machine is a genuinely useful piece of kit: it pre-stretches wires, measures tension transmission efficiency across eight pulley diameters and four rope types at 200 and 400 N, and adds a variable-length dynamic response measurement that prior work didn't cover. The efficiency curves are clean and monotonic – larger pulleys lose less, thicker wires lose more – and the numbers are consistent enough to be a design reference. The variable-length frequency response showing a ~5 dB low-frequency drop is a new observation that matters for any robot where wire length changes with posture.\n\nThe soft spot is the validation experiment. The claim that compensation reduces end-effector force error by 13% rests on two mean RMSEs: 10.9 N vs 9.5 N, five trials each, with no standard deviation, confidence interval, or test. Looking at Fig. 11, the two curves overlap for most of the ramp, so the difference could easily come from one phase of the motion or from trial noise. That's a real weakness in the paper's central quantitative claim. It doesn't undermine the utility of the efficiency table, but it does mean the robot control application isn't proven yet.\n\nA second, smaller issue is transferability: the per-pulley efficiency was measured at a fixed 90° wrap angle and two tension setpoints, then applied to a robot whose wires reverse direction and see varying tensions. The authors mention the wrap-angle limitation in the conclusion, but don't address direction reversal, and prior work (Choi et al.) shows that constant-efficiency models miss friction transients on reversal. So the compensation model is a reasonable first step, not a general method.\n\nThe paper is honest and clearly written. It cites the right prior work, reports its setup in enough detail to reproduce, and doesn't oversell beyond the data – except for the 13% claim, which is a simple overstatement given the missing error bars. I'd send this to peer review, but the referee should ask for per-trial RMSEs, variance, and ideally a wrap-angle sweep or at least a discussion of why direction reversal might be captured by the mean efficiency. It's a solid hardware contribution that deserves a serious look, and the dataset alone is likely to get cited.","headline":"Useful hardware and a useful efficiency dataset; the robot demo's 13% claim needs error bars before it carries weight.","tokens_in":9037,"tokens_out":2538,"would_cite":true,"duration_ms":22898,"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":"A wire testing machine that measures real pulley losses cuts a wire-driven robot's vertical force error by about 13 percent.","keywords":["wire-driven robots","tension transmission efficiency","passive pulley","force control","wire testing machine","coupled tendon-driven mechanism","pre-stretching","variable-length wire dynamics"],"falsifier":"Measure the actual tension just before and after every passive pulley on the robot while the end effector executes the same 0 to 40 N ramp, under both directions of wire motion, and compare each pulley's live efficiency with the single measured per-pulley efficiency. If live efficiencies differ substantially from the test-rig value, or if the compensation changes the sign of the tension correction on reversing wires, the 13 percent RMSE reduction should disappear or become negative.","tokens_in":8142,"feed_emoji":"🤖","tokens_out":5328,"duration_ms":44606,"temperature":0.7,"pith_summary":"This paper argues that the main obstacle to wider use of wire-driven robots, namely that flexible wires are lossy and hard to model, can be addressed by measuring the wires themselves. The authors built a Universal Wire Testing Machine that pre-stretches new wires, measures per-pulley tension transmission efficiency for eight pulley diameters and four wire types, and records tension dynamics while wire length changes. They then feed the measured per-pulley efficiency into the quadratic-program tension distributor of a real wire-driven robot. With compensation, the vertical end-effector force RMSE fell from 10.9 N to 9.5 N. The point is that a practical characterization rig, not a more complex model, is enough to make wire transmission more controllable.","feed_headline":"Wire tester data cuts robot force error by 13 percent","feed_subtitle":"Per-pulley efficiency fed into force control drops end-effector RMSE from 10.9 N to 9.5 N.","key_machinery":"The load-bearing object is the tension transmission efficiency matrix used inside the quadratic program. For each wire $i$ and joint $j$, the number of passive pulleys $N_{ij}$ converts the measured per-pulley efficiency $\\eta_p$ into $\\eta_{ij} = \\eta_p^{N_{ij}}$; the Hadamard product $\\eta \\odot G$ scales each wire's moment-arm contribution by the fraction of tension that survives the pulley routing, so the optimizer asks for more tension on wires that have lost more. The quadratic program minimizes $(\\tau_{\\text{ref}} + (\\eta \\odot G)^T T_{\\text{ref}})^T \\Lambda (\\tau_{\\text{ref}} + (\\eta \\odot G)^T T_{\\text{ref}}) + |T_{\\text{ref}}|^2$ subject to $T_{\\min} \\leq T_{\\text{ref}} \\leq T_{\\max}$, using the quadratic program formulation from prior work on redundant muscle tension. The testing machine itself supplies the measured $\\eta_p$ values, along with pre-stretching to remove initial wire stretch and a variable-length wire dynamics measurement system.","core_discovery":"The central discovery is that tension lost when a wire rounds a passive pulley can be turned from unmodeled error into a compensable quantity. The paper measures per-pulley efficiency as $\\eta_p = \\sqrt{T_{\\text{out}}/T_{\\text{in}}}$, then constructs an efficiency matrix $\\eta$ whose entries are $\\eta_p^{N_{ij}}$, where $N_{ij}$ is the number of passive pulleys wire $i$ passes through before reaching joint $j$. Replacing the muscle-length Jacobian $G$ by the elementwise product $\\eta \\odot G$ in the quadratic program that solves for target wire tensions (Eqs. 5 and 7) makes the optimizer request extra tension on wires that have lost more. In a robot using 1 mm Vectran rope and 12 mm pulleys, with up to seven passive pulley units per wire, this compensation reduced the vertical end-effector force RMSE from 10.9 N to 9.5 N, about 13 percent. The same experiments show transmission efficiency increases monotonically with pulley diameter and decreases with wire diameter, and that variable-length wires have different tension dynamics than constant-length wires.","pith_inferences":["A natural extension the paper leaves implicit is to make the efficiency matrix configuration-dependent: because wrap angles and the number of active pulley contacts change with robot posture, a lookup table indexed by joint angles could shrink the residual error further than the single 90-degree measurement.","The same per-pulley efficiency data could be inverted into a design tool: for a target force accuracy and available pulley diameters, a designer could choose wire diameter and routing to keep any wire's cumulative efficiency above a threshold.","Since prior work cited in the paper shows friction spikes when wires reverse direction, a direction-symmetric constant efficiency is the weakest point of the compensation; testing the compensated controller under cyclic reversing loads would show whether the 13 percent gain persists.","The variable-length dynamics measurement suggests a concrete testable claim: a tension-prediction model trained on length-varying data should beat a fixed-length viscoelastic model in the 2-10 Hz band where phase lag develops."],"forward_implications":["Force control on existing coupled wire-driven robots can be improved by roughly 13 percent simply by replacing the lossless muscle Jacobian with the measured-efficiency version, provided per-pulley efficiencies are known.","Increasing passive pulley diameter and decreasing wire diameter raises measured transmission efficiency monotonically, giving a direct design rule for low-friction wire routing.","Variable-length wires lose about 5 dB of tension-tracking magnitude at low frequencies and shift and attenuate resonance compared with constant-length wires, so dynamic models that ignore length variation will overestimate tension tracking.","Pre-stretching synthetic fiber rope removes initial plastic stretch, as demonstrated by an 8.2 m Dyneema rope elongating 0.43 m under 510 N for 12 hours, which should make strain-tension behavior more reproducible over the robot's life."],"supporting_citations":[{"why":"Establishes that tension loss in passive pulleys depends on tension, wrap angle, and number of sheaves and can be modeled as Coulomb friction, motivating the efficiency measurement.","marker":"[11]"},{"why":"Identifies internal fiber-fiber friction within the cable as the dominant source of tension loss over a pulley, supporting the measured efficiency values and the design conclusions.","marker":"[12]"},{"why":"Shows that constant-friction models miss tension behavior when wires reverse direction, providing the baseline limitation that the measured-efficiency compensation must address.","marker":"[16]"},{"why":"Supplies the quadratic-program formulation for computing redundant wire tensions under torque-error minimization and tension limits, which the paper extends with the efficiency matrix.","marker":"[19]"},{"why":"Documents creep and viscoelasticity of synthetic fiber ropes, motivating the pre-stretching function and the dynamic length-change measurements.","marker":"[15]"}],"fun_headline_variants":["Per-pulley efficiency data trims wire-robot force error 13%","Wire tester maps tension losses, cuts force error 13%","13% force-error cut with pulley efficiency matrix","Universal wire tester reduces robot force error to 9.5 N","Loss-compensated wire tension trims force error 13%"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The compensation assumes that the efficiency measured on the test rig at a 90-degree wrap angle and at 200 N and 400 N input tension applies unchanged to every pulley in the robot under actual tension, wire velocity, wrap angle, and direction of motion, even though the robot's wires reverse direction.","fun_headline_variants_meta":{"raw":{"variants":["Per-pulley efficiency data trims wire-robot force error 13%","Wire tester maps tension losses, cuts force error 13%","13% force-error cut with pulley efficiency matrix","Universal wire tester reduces robot force error to 9.5 N","Loss-compensated wire tension trims force error 13%"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000345,"raw_usage":{"total_tokens":1870,"prompt_tokens":901,"completion_tokens":969,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":517,"completion_tokens_details":{"reasoning_tokens":880}},"tokens_in":517,"tokens_out":969,"duration_ms":8630,"temperature":1.0,"reasoning_tokens":880,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-15T15:51:52.966321+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Measure the actual tension just before and after every passive pulley on the robot while the end effector executes the same 0 to 40 N ramp, under both directions of wire motion, and compare each pulley's live efficiency with the single measured per-pulley efficiency. If live efficiencies differ substantially from the test-rig value, or if the compensation changes the sign of the tension correction on reversing wires, the 13 percent RMSE reduction should disappear or become negative.","supporting_citations":[{"cited_title":"Measure- ment of the cable-pulley coulomb and viscous friction for a cable- driven surgical robotic system,","cited_arxiv_id":null,"evidence_quote":"Establishes that tension loss in passive pulleys depends on tension, wrap angle, and number of sheaves and can be modeled as Coulomb friction, motivating the efficiency measurement."},{"cited_title":"Measurement of the tension loss in a ca- ble traveling over a pulley, for low-speed applications,","cited_arxiv_id":null,"evidence_quote":"Identifies internal fiber-fiber friction within the cable as the dominant source of tension loss over a pulley, supporting the measured efficiency values and the design conclusions."},{"cited_title":"Tension analysis of a 6-degree- of-freedom cable-driven parallel robot considering dynamic pulley bearing friction,","cited_arxiv_id":null,"evidence_quote":"Shows that constant-friction models miss tension behavior when wires reverse direction, providing the baseline limitation that the measured-efficiency compensation must address."},{"cited_title":"A joint-space controller based on redundant muscle tension for multiple dof joints in musculoskeletal humanoids,","cited_arxiv_id":null,"evidence_quote":"Supplies the quadratic-program formulation for computing redundant wire tensions under torque-error minimization and tension limits, which the paper extends with the efficiency matrix."},{"cited_title":"Cable behavior influence on cable-driven parallel robots vibrations: experimental characterization and simulation,","cited_arxiv_id":null,"evidence_quote":"Documents creep and viscoelasticity of synthetic fiber ropes, motivating the pre-stretching function and the dynamic length-change measurements."}],"review_version":2}