{"id":"6f6532f8-4c24-4a61-a449-3a496081ca02","arxiv_id":"2502.03914","paper_version":1,"verdict":"CONDITIONAL","confidence":"HIGH","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":4,"one_line_summary":"A flexible FBG-based force sensor with integrated temperature compensation achieves sub-0.2 N accuracy and enables stable closed-loop grasping in a soft robotic gripper.","lead":"This paper describes a compact fiber-optic force sensor for soft robotic grippers that measures contact forces with high sensitivity and includes temperature compensation. The sensor was integrated into a soft gripper and shown to improve grip stability in automated pick-and-place tests.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Temperature compensation assumes identical temperature changes at FBG1 and FBG2; test only soaked both in the same water bath, so realistic thermal gradients remain unvalidated.","rationale":"The paper's central quantitative claims are the force RMSE values and the temperature-compensation RMSE of 0.01 N. The force calibration and load-cell comparison are internally consistent and support the force accuracy at constant temperature. The temperature-compensation result, however, rests entirely on the assumption that ΔT1 = ΔT2 (Eq. 5 in Section III). The authors explicitly state this assumption, but the experimental validation (Section IV.B) submerges the whole sensor in a single water bath, making the assumption true by construction. No test applies a thermal gradient between the two FBGs, and no test varies force during heating. The 10 mm separation and different housing materials mean that in a real gripping scenario—e.g., grasping a warm object or operating near sunlight—the two gratings may experience different temperatures. If the assumption fails, the compensation formula will not fully remove thermal drift, and the claimed accuracy is not preserved. This is a genuine limitation rather than a fatal flaw, because the sensor may still be accurate under uniform temperature environments, and the force-only claims are well supported. Hence the reader's CONDITIONAL verdict is appropriate, and I recommend no change to the verdict. The proposed Peltier-based test would directly measure the sensitivity of the compensated output to differential temperature and settle whether this concern lands.","tokens_in":11580,"tokens_out":10957,"duration_ms":100713,"concrete_test":"Mount the sensor with zero applied force in a rig where a Peltier element locally heats the bump region (FBG1) while the protective tube (FBG2) remains at ambient temperature. Record the compensated force estimate as the FBG1-FBG2 temperature difference is stepped from 0 to 5 °C. If the compensated force drifts by more than 0.01 N per degree of differential temperature, the ΔT1 = ΔT2 assumption is violated and the compensation claim only holds for uniform heating.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The sensor's temperature compensation model (Section III, Eq. 5) assumes ΔT1 = ΔT2, i.e., that the force-sensing FBG1 and the strain-free FBG2 experience identical temperature changes. This is an explicit but untested assumption: the two FBGs are 10 mm apart and housed in different structures (TPU chamber vs. Hytrel protective tube). The only validation (Section IV.B) immersed the entire sensor in a water bath, guaranteeing uniform temperature, and applied a constant 0.1 N force. Thus, neither differential thermal exposure nor any temperature dependence of the force-to-wavelength calibration (TPU modulus, etc.) has been experimentally verified. In a real gripper, local heating from the grasped object or ambient gradients could make ΔT1 ≠ ΔT2, and the compensation formula would then incompletely remove thermal drift, undermining the claimed force accuracy. This is load-bearing because 'effective temperature compensation' is a headline result.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper presents a compact fiber Bragg grating (FBG) based contact force sensor intended for robotic gripping systems. The sensor comprises a 3D-printed TPU casing with a bump and uvula structure, a dual FBG array, and a protective tube. FBG1 is strain-sensitive and located beneath the uvula, while FBG2 is strain-free and serves as a temperature reference. A second-order polynomial is used to calibrate wavelength shift to force, and a temperature compensation model uses the measured temperature sensitivity ratio between the two FBGs. The authors characterize the sensor through repeatability, hysteresis, calibration, load-cell comparison, and temperature compensation experiments. They then integrate the sensor with a soft grow-and-twine gripper on a UR5e arm and demonstrate that closed-loop PID force control improves object retention during pick-and-place tasks relative to open-loop grasping.","tokens_in":11749,"tokens_out":4058,"duration_ms":39045,"significance":"If the reported performance holds, the sensor offers a useful combination of compactness, high sensitivity (approximately 1169 pm/N), and reasonable accuracy (force RMSE 0.12 N against calibration and 0.14 N against a commercial load cell) for soft gripper force feedback. The experimental protocol is generally sound, with separate calibration and validation runs, explicit hysteresis and repeatability measurements, and a temperature compensation test. The integration demonstration is a clear strength: it provides a concrete application and shows that the sensor enables closed-loop force regulation that prevents object slippage. The main concerns are the incompleteness of the central derivation (missing equations) and the limited validation of the temperature compensation assumption under realistic thermal gradients. These are load-bearing issues for the paper's headline claims, but they are addressable with additional exposition or experiments.","major_comments":[{"comment":"In the submitted manuscript, the equations in Section III are blank placeholders; the actual formulas for wavelength shift, temperature compensation, and the force calculation are missing. This prevents the reader from verifying the derivation of Eq. (8), which is the core of the sensor's force measurement. The authors must include the full equations with all symbols defined. This is a load-bearing issue because the entire signal-processing chain rests on these expressions.","section":"Section III, Eqs. (2)-(8)"},{"comment":"The compensation model (Eq. 5) assumes ΔT1 = ΔT2, i.e., that the strain-sensing FBG1 and the strain-free FBG2 experience identical temperature changes. The validation test in Section IV.B submerges the entire sensor in a uniformly heated water bath, so both FBGs are at the same temperature by construction. No test is provided with differential heating between FBG1 in the TPU chamber and FBG2 in the protective tube, nor is the effect of local heating at the contact point examined. Consequently, the claim of effective temperature compensation is only supported for uniform temperature changes; its validity under realistic thermal gradients, where the two gratings could be at different temperatures, remains unverified. The authors should either add an experiment with localized heating or temper the claim and explicitly discuss this limitation.","section":"Section IV.B, temperature compensation test"},{"comment":"Only a single prototype is tested, and the force sensor is characterized exclusively under normal loading applied to the bump. The slot design is stated to 'minimize interference from small forces along the x- and z-axes,' but no off-axis or shear force data are reported. Since the sensor is mounted on a twining gripper where contact geometry and loading direction vary, the absence of off-axis characterization limits confidence in the force readings during actual grasping. This should be acknowledged as a limitation, and ideally quantified with at least a preliminary off-axis test.","section":"Section IV, force characterization"}],"minor_comments":[{"comment":"The figure numbering is inconsistent: the text repeatedly refers to Fig. 5a, 5b, and 5c for repeatability, hysteresis, and comparison results, but the captions show Fig. 4a-c for those results and Fig. 5a-b for the temperature tests. Similarly, the temperature results are referred to as Fig. 6a and 6b while the captions indicate Fig. 5. Please renumber the figures and cross-references consistently.","section":"Section IV.B, figure references"},{"comment":"The sentence 'This corresponds to a force measurement percentage error of approximately 2.56%' does not specify the denominator (full-scale, reading, or average). Please clarify how the percentage is computed.","section":"Section IV.B, percentage error statement"},{"comment":"The phrase 'the force sensor underwent a temperature change of 11 °C during this temperature compensation test' is clear, but later the text says the compensated force 'fluctuated around the saturated value of 0.1 N'; 'saturated' is not an appropriate term here. Consider replacing it with 'the value set by the tape tension' or 'the reference value'.","section":"Section IV.B, wording"},{"comment":"The reported sensitivity, 1169.04 pm/N, is given with four significant decimals while the measurement error is on the order of 0.1 N. Please round to a physically meaningful precision, e.g., 1169 pm/N.","section":"Abstract and Section IV.B, precision"},{"comment":"The PID gains are listed as Kp = 20, Ki = 0.05, and Kd = 0 without units. Since the controlled variable is force (in N) and the actuation is motor position or current, the units matter for reproducibility. Please specify the units or state that the gains are dimensionless.","section":"Section V, PID gains"}],"recommendation":"major_revision","confidential_remarks":"The manuscript is within the scope of the journal and reports a practical sensor with a thorough experimental protocol. The missing equations in Section III are the most serious issue; if this is an artifact of the PDF submission, the editor should verify the original file. The temperature compensation claim is somewhat overstated relative to the uniform-bath validation, and the single-prototype, normal-load-only characterization limits generality. I believe these issues are addressable with additional experiments or a revised discussion, so I recommend major revision rather than rejection."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Two things to know before you read it. The sensor geometry is actually new: a TPU casing with a bump and a uvula that presses a suspended fiber, plus a strain-free FBG in a protective tube for temperature compensation. That combination is not in their earlier ICRA paper or anywhere else in the cited literature. And the experimental effort is honest and mostly well executed: load-cell calibration, separate validation runs, hysteresis, and a pick-and-place demo with closed-loop PID.\n\nThe force-sensing results hold up. An RMSE of 0.12 N against calibration and 0.14 N against a commercial load cell are consistent, and the sensitivity of ~1169 pm/N is plausible for a suspended fiber under transverse load. The hysteresis of 4.83% is reported with a sensible discussion of material causes.\n\nThe soft spot is the temperature compensation. The model assumes ΔT1 = ΔT2, and the only validation test submerged the entire sensor in a water bath, guaranteeing uniform temperature. That is a real limitation: in a gripper, the force-sensing FBG near the contact point could see different thermal exposure than the strain-free FBG in the tube. The compensation would then only be approximate. This is not fatal, but the 'effective temperature compensation' headline should be scoped to uniform thermal conditions until a differential-temperature test is done.\n\nSmaller gaps: only one prototype, no off-axis or shear characterization, no shared data. These are worth asking for in revision but do not undermine the core claims for normal-force sensing.\n\nWho should read this: people working on soft gripper force feedback and FBG-based tactile sensing. It deserves a serious referee. I would send it to review with a request for a differential-temperature experiment and a second prototype. If those come back clean, it is a useful contribution.","headline":"Genuinely new sensor geometry with honest experiments, but the temperature-compensation claim is only proven under uniform temperature.","tokens_in":12277,"tokens_out":3153,"would_cite":true,"duration_ms":32152,"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 compact dual-grating fiber-optic sensor measures contact force on soft grippers with 0.12–0.14 N RMSE and cancels temperature drift.","keywords":["fiber Bragg grating","contact force sensing","soft robotic gripper","temperature compensation","closed-loop force control","slip prevention","tactile sensing"],"falsifier":"Run the sensor with FBG1 heated by a warm object at the contact bump while FBG2 stays at room temperature, keeping the mechanical load constant; if the compensated force output drifts by more than the reported 0.01 N RMSE or tracks the temperature difference, the equal-temperature assumption is violated and the compensation claim fails.","tokens_in":11395,"feed_emoji":"🤖","tokens_out":8657,"duration_ms":77053,"temperature":0.7,"pith_summary":"The paper reports a compact fiber-optic force sensor for soft robotic grippers and argues that it provides accurate, temperature-stable contact-force readings. The sensing element is a dual Fiber Bragg Grating (FBG) array: one grating is deformed by pressing a bump-and-uvula TPU casing, and the second, strain-free grating measures temperature alone so its signal can be subtracted. Bench tests give a sensitivity near 1169 pm/N, a force RMSE of 0.12 N against the calibration curve and 0.14 N against a commercial load cell, and a temperature-compensated force RMSE of 0.01 N over an 11 °C change. Integrated with a soft grow-and-twine gripper, the sensor's feedback let a PID controller hold contact force during pick-and-place, preventing the slips that occurred without control. The paper is establishing that this packaging solves the usual trade-off among sensitivity, fiber protection, and thermal drift for gripper force feedback.","feed_headline":"A thumb-size fiber-optic sensor steadies soft robotic grips","feed_subtitle":"Bench tests show 0.12 N accuracy, 0.01 N thermal drift, and pick-and-place runs with no dropped objects.","key_machinery":"The load-bearing mechanism is the dual-FBG array with asymmetric packaging. A Fiber Bragg Grating is a periodic refractive-index modulation in an optical fiber whose reflected wavelength shifts with strain and temperature; here FBG1 is suspended in a slot beneath the uvula, so pressing the external bump bends the fiber and changes its Bragg wavelength, while FBG2 sits strain-free in a protective tube and responds only to temperature. The compensation identity is $\\Delta\\lambda_{B,1} = K_{\\varepsilon 1}\\Delta\\varepsilon_1 + K_{T1}\\Delta T_1$ with $\\Delta T_1 = \\Delta T_2$, so the contact force is recovered from the two wavelength shifts through $N = A(\\Delta\\lambda_{B,1} - r\\Delta\\lambda_{B,2})^2 + B(\\Delta\\lambda_{B,1} - r\\Delta\\lambda_{B,2}) + C$. All reported performance numbers follow from this paired-grating subtraction plus the second-order calibration.","core_discovery":"The central discovery is that suspending an optical fiber beneath a 3D-printed TPU bump-and-uvula structure lets a single FBG measure transverse contact force directly, rather than relying on longitudinal stretching of a bonded fiber. A second FBG, glued strain-free inside a protective Hytrel tube, serves as a temperature reference; because the two gratings are 10 mm apart, the paper assumes they see the same temperature change and removes the thermal term using an experimentally measured sensitivity ratio r = KT1/KT2 = 2.356. The relation between applied force and the temperature-corrected wavelength shift is fitted as a second-order polynomial, reflecting the nonlinear bending-and-stretching of a beam fixed at both ends and loaded at the center. The paper claims this sensor is repeatable over 0–4.69 N with 4.83% maximum hysteresis, and that closed-loop PID grip control using its readings kept a 351 g crimper, a 419 g hammer, and a 222 g bottle from slipping during automated transfer, while uncontrolled grasps dropped all three.","pith_inferences":["Extension, not paper claim: because the equal-temperature assumption is untested under the thermal asymmetry of real contact, a field deployment would need either a third grating to measure the gradient or calibration of r against temperature difference.","Extension, not paper claim: the sensor's roughly 0.86 mN resolution and fast FBG response imply it could detect incipient slip events before gross slip, but the paper only demonstrates post-slip force recovery, not prediction.","Extension, not paper claim: the 4.69 N range and 4.83% hysteresis bound the usefulness for precision assembly; stiffer casing materials or larger chamber heights could extend the range at the cost of sensitivity, a trade-off the paper notes only as future work."],"forward_implications":["A 12×12×4 mm FBG sensor can close the force-control loop on a soft gripper without exposed fiber, reducing fragility concerns.","Without feedback all three test objects slipped during transfer; with feedback, none slipped, so force feedback from this sensor directly improves grasp retention.","Temperature compensation cuts thermal force error from 0.23 N RMSE to 0.01 N RMSE over an 11 °C change, making the sensor usable outside temperature-stable lab conditions.","The calibration is repeatable enough (R² = 0.99) for a force feedback signal, with about 0.86 mN resolution given the interrogator's 1 pm resolution.","The design's small size and protective tube point toward arrays and multimodal vision-plus-force systems for harvesting, logistics, and prosthetics, as the paper's conclusion sketches."],"supporting_citations":[{"why":"Supplies the Bragg condition linking wavelength to grating pitch, the basis for converting wavelength shifts into strain and temperature measurements.","marker":"[20]"},{"why":"Gives the wavelength-shift law separating thermal and strain contributions, on which the dual-grating subtraction is built.","marker":"[21]"},{"why":"Provides the large-deformation beam model that justifies the second-order polynomial calibration between applied force and wavelength shift.","marker":"[22]"},{"why":"Supplies the grow-and-twine gripper hardware used in the closed-loop pick-and-place integration and grasp-stability tests.","marker":"[18]"}],"fun_headline_variants":["Optical fiber sensor gives soft grippers a sense of touch","Tiny fiber sensor lets soft robots feel grip force precisely","FBG sensor measures contact force with 0.12 N accuracy","Thermally compensated fiber sensor steadies soft robotic grips","Closed-loop force control with fiber sensor prevents slips"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The compensation assumes that FBG1 and FBG2 experience identical temperature changes at all times, even though they are 10 mm apart and housed in different structures, and the validation heated both together in a water bath rather than under the asymmetric heating of a real grip.","fun_headline_variants_meta":{"raw":{"variants":["Optical fiber sensor gives soft grippers a sense of touch","Tiny fiber sensor lets soft robots feel grip force precisely","FBG sensor measures contact force with 0.12 N accuracy","Thermally compensated fiber sensor steadies soft robotic grips","Closed-loop force control with fiber sensor prevents slips"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000677,"raw_usage":{"total_tokens":3150,"prompt_tokens":1089,"completion_tokens":2061,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":705,"completion_tokens_details":{"reasoning_tokens":1979}},"tokens_in":705,"tokens_out":2061,"duration_ms":15801,"temperature":1.0,"reasoning_tokens":1979,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-09T00:14:22.643484+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Run the sensor with FBG1 heated by a warm object at the contact bump while FBG2 stays at room temperature, keeping the mechanical load constant; if the compensated force output drifts by more than the reported 0.01 N RMSE or tracks the temperature difference, the equal-temperature assumption is violated and the compensation claim fails.","supporting_citations":[{"cited_title":"Force Sensing With 1 mm Fiber Bragg Gratings for Flexible Endoscopic Surgical Robots,","cited_arxiv_id":null,"evidence_quote":"Supplies the Bragg condition linking wavelength to grating pitch, the basis for converting wavelength shifts into strain and temperature measurements."},{"cited_title":"A Three-Axial Force Sensor Based on Fiber Bragg Gratings for Surgical Robots,","cited_arxiv_id":null,"evidence_quote":"Gives the wavelength-shift law separating thermal and strain contributions, on which the dual-grating subtraction is built."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Provides the large-deformation beam model that justifies the second-order polynomial calibration between applied force and wavelength shift."},{"cited_title":"Smart Grow-and-Twine Gripper for Vegetable Harvesting in Vertical Farms,","cited_arxiv_id":null,"evidence_quote":"Supplies the grow-and-twine gripper hardware used in the closed-loop pick-and-place integration and grasp-stability tests."}],"review_version":1}