{"id":"8a328449-841d-46ac-8b06-4af64d382adf","arxiv_id":"1908.02689","paper_version":2,"verdict":"CONDITIONAL","confidence":"HIGH","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":2,"one_line_summary":"A new cable-driven robotic interface, NED, accurately identifies lower limb joint mechanical impedance in an upright posture without constraining joint motion.","lead":"This paper introduces a cable-driven robot that can measure the mechanical stiffness, damping, and inertia of the hip, knee, and ankle while a person stands or sits naturally. The device is validated on a dummy leg, springs, and one human volunteer, showing promise for studying leg mechanics in rehabilitation robotics.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The cable transmission's measured stiffness (~500 N/m) is comparable to the validation springs and to the joint being identified, so the assumed rigid mapping from motor motion to leg motion is not established and can bias the reported impedance estimates.","rationale":"The paper's headline claim is that NED can accurately identify single-joint neuromechanics, and the strongest validation evidence is the dummy-leg inertia match and the spring stiffness test. The reader identified the load-bearing assumption as effective rigidity of the cable transmission; this is also the point where I find the quantitative support weakest. Section III-C explicitly measures the cable as a compliant second-order system with stiffness of only about 500 N/m, while the validation springs in Section IV-B have stiffnesses of roughly 200-600 N/m. In any series arrangement, two comparable compliances produce a combined stiffness substantially below either one, so the reported agreement in Fig. 8d cannot by itself show that the spring's true stiffness was recovered. The same issue applies to the human pilot, where no gold-standard joint torque measurement or endpoint motion capture is available. This is not a claim that the device is necessarily inaccurate; it is a claim that the paper's evidence does not yet rule out a systematic bias of the same order as the quantities being measured. The proposed tests—recomputing the series prediction and using a much stiffer spring with an independent endpoint sensor—would settle the concern directly. Since the reader's conditional verdict already accounts for this uncertainty, I do not recommend changing the verdict.","tokens_in":13519,"tokens_out":5736,"duration_ms":72486,"concrete_test":"Report the independently measured spring constant Ks used in Section IV-B and the cable stiffness Kx from Section III-C in matching units, then recompute the series prediction Kx*Ks/(Kx+Ks) and compare it with the NED estimates in Fig. 8d. If the estimates match the series prediction rather than Ks, the rigid-transmission assumption fails. In addition, run one spring trial with Ks near 5000 N/m while recording true foot-fixture displacement with an optical marker; if the estimated stiffness saturates near the cable stiffness or differs from the marker-based estimate, the motor-encoder mapping in Eq. (4) is the source of bias.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central claim of accurate impedance identification rests on Eq. (4), which treats motor-side displacement as the leg's displacement and fits I, B, K to that signal. Section IV-A justifies this by stating that 'the cable dynamics in series with the leg could be neglected', but Section III-C itself identifies the cable as a second-order system with stiffness Kx near 500 N/m (0.5 N/mm; Fig. 6b) and poles at 1.6 and 17.2 Hz. This is the same order of magnitude as the spring stiffnesses validated in Section IV-B (0.2-0.6 N/mm, Fig. 8d) and, after multiplying by L^2 for a 0.7 m leg, corresponds to about 245 Nm/rad of series rotational compliance, a value comparable to the joint impedances being estimated. If the spring in Fig. 8a is attached at the cable endpoint, the stiffness seen by the motor is the series combination Kx*Ks/(Kx+Ks), which for Kx=Ks=500 N/m is about 250 N/m, not 500 N/m; the paper does not report an independently measured spring constant or a series-compliance correction. The same unmodeled compliance shifts the apparent inertia and damping in Eq. (4) whenever the perturbation contains energy near the 1.6 Hz cable pole, and the 150 ms plateau used in the human pilot has such spectral content. No optical endpoint measurement or motion-capture comparison is reported, so the rigid-transmission assumption is insufficiently supported for the quantitative accuracy claim.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper describes NED, a cable-driven endpoint-based robotic interface for identifying single-joint neuromechanics (inertia, viscosity, stiffness) of the hip, knee, and ankle in an upright, unconstrained posture. The authors characterize the device's kinematics, force measurement sensitivity, and cable transmission dynamics, then validate impedance identification on a rigid dummy leg (inertia compared with CAD) and on linear springs of known stiffness, and finally present a single-subject pilot for hip impedance. The central claim is that the device can accurately identify the mechanical impedance of a limb, with the identification model given in Eq. (4): Δτ = IΔθ¨ + BΔθ˙ + KΔθ.","tokens_in":13850,"tokens_out":2668,"duration_ms":30622,"significance":"If the accuracy claim holds, NED would be a useful addition to lower-limb neuromechanics research, because existing dynamometers constrain joints and exoskeletons have limited rigidity; the ability to test hip, knee, and ankle in a natural upright posture without joint constraints fills a practical gap. The paper provides careful mechanical characterization, multiple validation conditions including a dummy leg and springs, and a human pilot, which is more than many device papers offer. The validation targets are independent (CAD inertia, known spring constants, Winter's anatomical model), which is a strength. However, the central claim of accurate impedance identification rests on the assumption that the cable transmission is effectively rigid, and the manuscript's own cable identification data challenge that assumption. The result is therefore promising but not yet established.","major_comments":[{"comment":"The identification model Eq. (4) treats the motor-encoder displacement as the leg's angular displacement, neglecting the series cable compliance. The paper's own cable characterization in Section III-C reports a cable stiffness around 500 N/m (Fig. 6b) with poles at 1.6 and 17.2 Hz. This stiffness is the same order of magnitude as the spring constants validated in Section IV-B and, after conversion to a rotational stiffness for a ~0.7 m leg, is comparable to the joint impedances being estimated. With a series compliance Kx in series with the leg impedance, the stiffness seen by the motor is not the leg stiffness alone; for Kx = Ks, the apparent stiffness is halved. No independent measurement of the leg endpoint displacement (e.g., optical tracking) and no series-compliance correction are reported, so the quantitative accuracy claim in the abstract is not supported as it stands.","section":"Section IV-A, Eq. (4)"},{"comment":"The spring stiffness validation in Eq. (5) also uses motor-side displacement ΔX and measured force difference ΔF. If the spring is attached beyond the cable and the load-cells measure the cable tension, then the system identified includes the cable stiffness in series with the spring, and the motor-side stiffness is Kx*Ks/(Kx+Ks) rather than Ks. The paper does not state where the load-cells are located relative to the spring or the cable, does not report an independently measured spring constant, and does not compare motor displacement with spring endpoint displacement. Without these details, the claim that the estimated stiffness is 'similar to the spring constant' does not confirm that the device measures the spring's actual stiffness.","section":"Section IV-B, Eq. (5)"},{"comment":"The human pilot applies the same rigid-transmission assumption when fitting Eq. (4) to motor-encoder data. The perturbation uses a 150 ms plateau, which contains spectral content near the 1.6 Hz cable pole identified in Section III-C. Since no motion capture or other endpoint measurement is used, the reported hip inertia, viscosity, and stiffness values may contain contributions from the cable dynamics. The small variance and the agreement of the inertia estimate with Winter's model are encouraging, but they do not resolve the transmission-identification confound. A definitive validation would require either endpoint motion measurement or inclusion of the identified cable dynamics in the identification model.","section":"Section IV-C, Fig. 9"}],"minor_comments":[{"comment":"In Eq. (2), ρ_m is described as the motor pulley diameter, but the relation ˙x = ρ_m ˙θ_m is correct only if ρ_m is the pulley radius. Please clarify which quantity is used.","section":"Section III-A, Eq. (2)"},{"comment":"The pilot subject is described as female, but the text later says 'asked to support his body weight' and 'his leg'; please make the pronouns consistent.","section":"Section IV-C"},{"comment":"The identified cable stiffness values in Fig. 6b appear to vary considerably across speeds (roughly 0 to 600 N/m), yet the text states 'low variance' without reporting quantitative variance measures. Please provide error bars or confidence intervals for the identified parameters.","section":"Section III-C, Fig. 6b"},{"comment":"The text notes that adding a motion capture system 'would enable to measure hip joint rotations directly and could improve the leg displacement measurement relative to the current estimation from the motor encoder.' This is a relevant limitation of the current study and should be discussed explicitly in Section V, not only in the hardware description.","section":"Section II-C"}],"recommendation":"major_revision","confidential_remarks":"The series-compliance issue is the crux: the paper's own cable stiffness data contradict the rigid-transmission assumption used in the identification, and the spring validation may itself be affected by the same confound. The revision should provide either endpoint displacement measurements or an explicit series-compliance correction in the identification model; otherwise the central accuracy claim is not established. The paper's scope is appropriate for IEEE TBME, and the authors have access to the necessary equipment to perform the additional validation. I also note that some key nonlinearity and temporal-dependency details are relegated to a PhD thesis [32]; for a journal paper, the essential information should be self-contained."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Couple of things to know. The hardware concept is genuinely new and worth knowing about: a cable-driven, endpoint-based platform that can perturb hip, knee, or ankle in an upright posture without constraining the joint. That fills a real gap. The paper also does honest system identification work on the cable itself. But the central validation argument has a soft spot that runs straight through it. Their own identified cable stiffness is around 500 N/m (Section III-C, Fig. 6), which is not rigid in any meaningful sense. Yet Section IV-A just asserts that \"the cable dynamics in series with the leg could be neglected.\" That assumption is doing load-bearing work. If the cable compliance is in series with the leg, then using the motor encoder as the leg's displacement in Eqs. (4) and (5) biases all three impedance parameters. The stress-test note makes the calculation: with Kx ~500 N/m and leg length 0.7 m, the rotational series compliance is ~245 Nm/rad, the same order as the joint stiffnesses being estimated. The spring validation does not refute this because the paper never reports the actual spring constants—so the estimated stiffness could be the series combination, not the spring alone. The dummy leg inertia matches CAD, which is a nice check, but a matching inertia does not rule out a series compliance bias; it just means the fitted inertia happens to come out right, and the spring test is the place to catch it if ground truth were reported. The human pilot is one subject and only hip, with no gold-standard comparison. Knee and ankle are claimed but not validated at all. So the paper is a solid hardware contribution with a validation plan that is not yet airtight. The math and the citations look fine. The cable identification itself is a good piece of work. I would send it to peer review, but I would insist the authors either measure the endpoint position directly or correct for the identified cable compliance before claiming accurate impedance identification. For a reader it is a useful paper to know about, but I would not use the reported impedance numbers yet.","headline":"Genuinely new cable-driven lower-limb platform; the validation claim is undercut by the paper's own cable compliance identification, so treat the impedance numbers with caution.","tokens_in":14360,"tokens_out":6218,"would_cite":false,"duration_ms":66217,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"A cable-driven robotic interface can identify hip, knee, and ankle neuromechanics in an upright posture without constraining the joints.","keywords":["cable-driven robot","neuromechanics identification","joint impedance","hip joint stiffness","lower limb","position perturbation","system identification","endpoint-based interface"],"falsifier":"Instrument the leg endpoint with a motion capture marker and repeat the spring and dummy-leg identifications; if the endpoint displacement differs from the motor-encoder displacement by more than the paper's 5% error bound, the rigid-cable assumption is violated. Alternatively, fit the same data with the measured cable stiffness placed in series with the identified joint and check whether the estimated stiffness shifts by more than the reported variance.","tokens_in":13329,"feed_emoji":"🦵","tokens_out":6734,"duration_ms":65404,"temperature":0.7,"pith_summary":"The paper sets out to show that one cable-driven, endpoint-based robotic interface can identify the mechanical impedance—inertia, viscosity, and stiffness—of the hip, knee, and ankle in a natural upright posture, without locking the joint into a fixed axis. Existing motor-driven dynamometers constrain joint motion and cannot easily reach the hip under weight-bearing conditions, while exoskeletons add structural vibrations and constrain the joints. The authors build the Neuromechanics Evaluation Device (NED), in which a powerful actuator fixed outside the workspace drives a pretensioned cable attached at the ankle, and validate it with an 18 kg dummy leg and springs of known stiffness. A pilot on one human subject produces reproducible hip impedance estimates whose inertia matches an anatomical model. If the claim holds, researchers could measure lower-limb neuromechanics in the postures that matter for standing and walking, using a single device.","feed_headline":"Cable-driven rig reads hip, knee, and ankle stiffness","feed_subtitle":"Pretensioned cables estimate inertia, viscosity, and stiffness of leg joints without constraining them.","key_machinery":"The central object is NED itself, a cable-driven endpoint interface whose actuator sits on the floor, connected by a pretensioned steel cable loop to a foot fixture with two load cells; lockable pulleys and an open seat let the same rig address hip, knee, or ankle. The load-bearing kinematics is the planar mapping $\\dot{\\theta} = \\rho_m \\dot{\\theta}_m / L$ of Eq. (2), which turns motor-pulley motion into joint rotation, and the impedance identity of Eq. (4), which turns differential cable force $(F_1-F_2)L$ and joint angle into $I$, $B$, and $K$. The capstan equation $T_{\\text{load}} = T_{\\text{hold}} e^{\\mu\\phi}$ sizes the cable wrap to prevent slippage, and a 200 N pretension limits sagging-induced force error to below 2.5%. A smooth ramp-and-hold perturbation with a constant-position plateau lets Eq. (5), $\\Delta\\tau = K\\Delta\\theta$, read stiffness directly at the plateau because acceleration and velocity vanish there.","core_discovery":"The core claim is that, once the cable dynamics in series with the leg are neglected, the leg's motion obeys a linear second-order joint model $\\Delta\\tau = I\\Delta\\ddot{\\theta} + B\\Delta\\dot{\\theta} + K\\Delta\\theta$, with $\\Delta\\tau = (F_1 - F_2)L$ read from two load cells, so a least-squares fit of a fast position perturbation returns the joint's inertia $I$, viscosity $B$, and stiffness $K$. Mechanical characterization shows the interface itself behaves as a stiff second-order system with stiffness above 500 N/m and low viscosity. Identification of a rigid dummy leg recovers its CAD-computed inertia of $1.84\\,\\mathrm{kg\\,m^2}$, and spring tests recover known stiffness values, with accuracy improving at larger perturbation amplitudes. Applying the same perturbation on one relaxed human subject yields hip impedance estimates with small variance and inertia close to an anatomical prediction, supporting the claim that NED can identify single-joint neuromechanics in an upright posture without constraining the joint.","pith_inferences":["A direct way to test the rigid-cable assumption is to include the identified cable stiffness as a series spring in the impedance model and see whether estimated joint stiffness changes by a detectable amount.","Adding motion capture of the actual joint angle, rather than the motor encoder, would check the planar rigid-leg kinematic mapping and could extend the method to multi-joint perturbations.","The same endpoint cable concept could scale to other joints or to patient populations, since the independent support removes the need for the subject to balance during measurement.","The pilot compares only inertia to an anatomical model; a same-subject comparison of stiffness and viscosity against an independent method would confirm human accuracy."],"forward_implications":["One device can map hip, knee, and ankle neuromechanics by relocating pulleys and adjusting the seat, with the same actuator and control chain.","A single perturbation protocol yields all three impedance parameters: stiffness from the plateau, inertia and viscosity from the transient fit.","The interface supports isometric, isokinetic, and brief high-speed perturbations up to 750 mm/s, covering protocols used for reflex and spasticity assessment.","Because body weight is carried by an independent structure and the joint is not mechanically fixed, measured mechanics correspond to upright, natural postures.","Hip viscoelasticity, previously accessible mainly through less accurate multi-joint perturbations, can now be probed with a precise single-joint position displacement."],"supporting_citations":[{"why":"Supplies the multi-joint hip and knee dynamics estimates against which the human viscoelasticity values are compared in order of magnitude.","marker":"[17]"},{"why":"Provides lower-limb stair-climbing joint torques used to set the required standstill torque of the actuator.","marker":"[26]"},{"why":"Anatomical model used to compute the pilot subject's expected leg inertia for comparison with the fitted value.","marker":"[28]"},{"why":"Establishes that pretensioned cable transmission provides low inertia and backlash-free force transmission, motivating the cable design.","marker":"[31]"},{"why":"Contains the actuator selection process, NED component details, and identification of cable nonlinearities.","marker":"[32]"},{"why":"Supplies the smooth ramp-and-hold endpoint stiffness method used for spring stiffness estimation.","marker":"[33]"},{"why":"Defines the fast ankle perturbation amplitude and pulse width used as a functional requirement for dynamic perturbation capability.","marker":"[18]"},{"why":"Complements the reflex-stiffness protocol and reinforces the need for accurate fast perturbations.","marker":"[19]"}],"fun_headline_variants":["Cable robot pinpoints lower-limb joint impedance","Cable-driven interface measures hip, knee, ankle stiffness","No-constraint cable rig quantifies leg neuromechanics","Cable system identifies joint inertia, viscosity, and stiffness"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The identification treats the cable transmission as effectively rigid and the leg as a rigid straight segment rotating in the sagittal plane, so the motor-side displacement is taken to equal the joint rotation; if the measured cable stiffness (around 500 N/m, comparable to the validation springs) cannot be neglected, the estimated joint stiffness would be biased.","fun_headline_variants_meta":{"raw":{"variants":["Cable robot pinpoints lower-limb joint impedance","Cable-driven interface measures hip, knee, ankle stiffness","No-constraint cable rig quantifies leg neuromechanics","Cable system identifies joint inertia, viscosity, and stiffness"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000159,"raw_usage":{"total_tokens":1192,"prompt_tokens":873,"completion_tokens":319,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":489,"completion_tokens_details":{"reasoning_tokens":252}},"tokens_in":489,"tokens_out":319,"duration_ms":4015,"temperature":1.0,"reasoning_tokens":252,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T14:38:04.892716+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Instrument the leg endpoint with a motion capture marker and repeat the spring and dummy-leg identifications; if the endpoint displacement differs from the motor-encoder displacement by more than the paper's 5% error bound, the rigid-cable assumption is violated. Alternatively, fit the same data with the measured cable stiffness placed in series with the identified joint and check whether the estimated stiffness shifts by more than the reported variance.","supporting_citations":[{"cited_title":"A study of lower-limb mechanics during stair-climbing","cited_arxiv_id":null,"evidence_quote":"Provides lower-limb stair-climbing joint torques used to set the required standstill torque of the actuator."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Anatomical model used to compute the pilot subject's expected leg inertia for comparison with the fitted value."},{"cited_title":"The effect of transmission design on force- controlled manipulator performance,","cited_arxiv_id":null,"evidence_quote":"Establishes that pretensioned cable transmission provides low inertia and backlash-free force transmission, motivating the cable design."},{"cited_title":"Development of the neuromechanics evaluation device (NED) for subject-specific lower limb modelling of spinal cord injury,","cited_arxiv_id":null,"evidence_quote":"Contains the actuator selection process, NED component details, and identification of cable nonlinearities."},{"cited_title":"Intrinsic and reflex stiffness in normal and spastic, spinal cord injured subjects,","cited_arxiv_id":null,"evidence_quote":"Complements the reflex-stiffness protocol and reinforces the need for accurate fast perturbations."}],"review_version":1}