{"id":"998db68c-dbfc-4423-acc1-ddc2e256189d","arxiv_id":"2502.03974","paper_version":1,"verdict":"REJECT","confidence":"MODERATE","novelty_score":3.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":3,"one_line_summary":"A lead-lag judgment with acceleration compensation is proposed for spatiotemporal trajectory tracking, with a simulated 1.21 m error on a highway ramp.","lead":"The paper proposes a lead-lag judgment mechanism that measures how far ahead or behind a vehicle is from a desired trajectory over time, then applies acceleration corrections to reduce that error. It reports real-vehicle experiments and a highway simulation, claiming tracking errors around one meter during ramp merging.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The 1.21 m simulation result is produced by an idealized curvature-following model with no baseline ablation and an under-specified compensation law (Eq. 7), so it does not validate the proposed mechanism.","rationale":"The reader's verdict REJECT identifies a real gap: the simulation uses an idealized vehicle response, and the real-vehicle experiments do not include the proposed compensation. My stress-test focuses on an even more specific and load-bearing defect: the Section 4 simulation does not contain a falsifiable vehicle model at all, Eq. (7) has an undefined parameter T and undisclosed gains, and no baseline/ablation run is reported. This means the ~1.21 m error is not attributable to the lead-lag compensation with any confidence. The reader's weakest_assumption (idealized vehicle response) is closely related but slightly broader; I would sharpen it to 'the simulation's trajectory generator, not the proposed controller, may be producing the reported accuracy.' This supports rather than changes the REJECT verdict, so the recommendation is UNCHANGED. I am not claiming fraud or intentional omission; the paper simply does not provide enough information to reproduce the simulation or distinguish the mechanism's contribution from the background model. Rule 8 is satisfied because the concrete ablation and model-fidelity test would settle whether the concern lands: if the no-compensation run shows similar error, the central claim falls; if the lagged-model run preserves the sub-1.5 m error and the ablation shows a large degradation, the concern would be refuted and the claim would be substantially strengthened.","tokens_in":10108,"tokens_out":3259,"duration_ms":33952,"concrete_test":"Re-run the Section 4 simulation in three configurations with the same input highway geometry: (i) the paper's described setup; (ii) the same setup with acceleration compensation removed (a_compensation = 0); and (iii) the same setup with a first-order actuator lag on longitudinal acceleration (e.g., 5 rad/s bandwidth) and a curvature-dependent lateral tracking error model. Report the maximum spatiotemporal error for each configuration. If configuration (ii) gives a maximum error comparable to 1.21 m, the proposed mechanism is not responsible for the reported performance; if configuration (iii) significantly increases the error, the idealized model is load-bearing and the real-vehicle transferability claim fails.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The paper's central claim — that the lead-lag judgment plus real-time acceleration compensation is validated by achieving a ~1.21 m spatiotemporal error at 100 km/h — rests entirely on the Section 4 simulation. That simulation is not a vehicle model: the text states that 'the vehicle speed and heading angle were adaptively adjusted in accordance with the curvature characteristics of the original path.' This is essentially a perfect-following kinematic generator, not a model with actuator lag, tire dynamics, or lateral-longitudinal coupling. The acceleration compensation law in Eq. (7) contains an undefined parameter T and no numerical gains or simulation time step are disclosed, so the reported error cannot be independently reproduced. Critically, no ablation is reported: there is no simulation run with the compensation disabled, and no comparison to a baseline controller. The real-vehicle experiments in Section 3 used only LQR plus dual PID and the lead-lag criterion only for post-processing, not for control. Therefore the measured 1.21 m error could be dominated by the assumed trajectory-generation rule rather than by the proposed mechanism, and Conclusion item 2's claim of 'validating its effectiveness' is unsupported.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper proposes a lead-lag judgment mechanism for spatiotemporal trajectory tracking: using the dot product between the vehicle-to-target position vector and the target-path tangent (Eqs. (1)-(4)) to classify the tracking point as ahead or behind, a companion dot product with the normal (Eqs. (5)-(6)) for lateral side, and an acceleration-compensation law (Eq. (7)) with a ±0.5 m deadband that adds positive or negative acceleration when the longitudinal deviation exceeds the threshold. The authors report real-vehicle experiments at 20-30 km/h on straight, curved, and combined tracks using an LQR+dual-PID controller, with the lead-lag criterion applied only in post-processing to quantify speed, heading, lateral, and advance/delay errors. They then present a highway simulation based on the S81 Huili-Luquan expressway alignment with a desired speed of 100 km/h, reporting a maximum spatiotemporal error of about 1.21 m (Fig. 5(f)), which is taken to validate the compensation mechanism. The paper claims implications for safe spatiotemporal distances in preemptive highway ramp merging.","tokens_in":10317,"tokens_out":7780,"duration_ms":69272,"significance":"If the central claim were supported, the contribution would be of genuine interest: temporal (as opposed to purely spatial) trajectory tracking is under-served in the CAV merging literature, and a cheap, interpretable lead/lag classifier could be useful for diagnostic and supervisory layers. The dot-product criterion itself is clearly and compactly defined, and the real-vehicle data set provides a reasonable empirical description of how speed, heading, and lateral errors behave on straight and curved sections at low speed. Those are strengths. However, the paper's headline result — 'validating its effectiveness' of the compensation mechanism at 100 km/h (Conclusion item 2) — is not established. The validating simulation is generated by a kinematic rule that adjusts the vehicle speed and heading to the path curvature (Section 4), the compensation law has undisclosed parameters (Eq. (7)), there is no baseline or ablation, and the field experiments never run the proposed compensation online.","major_comments":[{"comment":"The central validation claim rests on a simulation that does not model vehicle dynamics. Section 4 states that 'the vehicle speed and heading angle were adaptively adjusted in accordance with the curvature characteristics of the original path,' which describes a kinematic trajectory generator rather than a vehicle with actuator lag, tire dynamics, or lateral-longitudinal coupling. The reported maximum spatiotemporal error of approximately 1.21 m (Fig. 5(f)) is therefore a property of that generator under the proposed compensation, and the claim in Conclusion item 2 that the algorithm 'validated its effectiveness' is unsupported: the tracking trajectory and the error being measured are both produced by the same mechanism, making the validation circular. A faithful test needs a vehicle model, or real closed-loop experiments, that is not already slaved to the target path's curvature.","section":"Section 4; Conclusion item 2"},{"comment":"The compensation law in Eq. (7) contains two parameters, ω and T, that are never defined or given numerical values anywhere in the paper; T appears in the equation but is not even identified as the intended time constant, and no tuning procedure, initial conditions, or simulation time step are disclosed. Because Eq. (7) is the proposed mechanism whose effectiveness is the paper's headline claim, the 1.21 m result cannot be reproduced, and it cannot be determined whether the reported error is sensitive to these parameters or simply reflects a favorable choice. This is a load-bearing omission for a validation claim; at minimum, a table of all parameters and a sensitivity or ablation study over ω and T is required.","section":"Section 2.2, Eq. (7)"},{"comment":"No baseline or ablation isolates the effect of the proposed mechanism. Section 4 reports only the compensated run; there is no simulation with the compensation disabled and no comparison with the LQR+dual-PID controller used in the field experiments. Conversely, the real-vehicle experiments in Section 3 never activate the lead-lag compensation online — the Abstract states that the tracking data 'was subsequent processed using the lead-lag judgment mechanism,' i.e., the criterion is applied only in post-processing. Consequently, no part of the paper demonstrates empirically or in simulation that the compensation reduces error relative to an existing controller; a cross-comparison (same scenario, with and without compensation) is the minimum needed to support the Abstract's claim of 'significantly enhanced' accuracy.","section":"Section 3 vs. Section 4"},{"comment":"The claims about highway ramp merging are not supported by the presented scenarios. The simulation in Section 4 is a single vehicle tracking a single path on a highway alignment, and the field tests are single-vehicle, low-speed tracking runs. There is no ramp-merging scenario with multiple vehicles, conflicts, or merge-sequence constraints anywhere in the paper, yet the Abstract concludes that 'reasonable spatiotemporal distance is given during the preemptive merging process on highway ramps' and the Conclusion states the algorithm was designed to 'establish safe spatiotemporal distances for highway ramp merging.' These statements overstate the scope of what was tested; they should either be removed or supported by a multi-vehicle merging simulation.","section":"Abstract; Section 4; Conclusion"}],"minor_comments":[{"comment":"The manuscript needs a careful language edit: examples include 'was subsequent processed' (Abstract), 'an lead/lag judgment' (Conclusion item 1), and the pervasive garbled spacing in the displayed equations and figure captions, which obscures the derivations.","section":"Throughout"},{"comment":"The experimental platform is not described adequately for replication: vehicle type, positioning/sensor system, the LQR and PID gains, and sampling times are all missing, which prevents independent verification of the reported error measurements.","section":"Section 3"},{"comment":"The symbol T is used both in '0 ≤ t ≤ T' in Section 2.1 and in the compensation law of Eq. (7); the two appear unrelated and neither is defined. Also, the limit notation in Eqs. (2)-(3) (written as '0t → 0') is unclear; presumably a limit Δt → 0 is intended and should be rewritten.","section":"Section 2.1 and Eq. (7)"},{"comment":"Several subplots lack axis labels and units in the captions, and the speed-error intervals quoted in the text (e.g., '[−0.1, 0.5] m/s') should be stated with units and made consistent with Fig. 5(c). The captions of Figs. 2-4 also label their subplots inconsistently.","section":"Figures 2-5"},{"comment":"The reference list has malformed entries: [30], [31], and [33] are self-citations of arXiv preprints with broken identifiers (e.g., '241101918' missing the dot), and [22] lacks volume and page information. The authors' own preprints should be cited in final published form where available.","section":"References"},{"comment":"The Abstract states that tracking errors were 'maintained in an acceptable range' without defining the acceptance criterion; specify the quantitative threshold used to judge acceptability.","section":"Abstract"}],"recommendation":"reject","confidential_remarks":"The paper's central validation is effectively circular: the only test of the compensation mechanism is a kinematic simulation whose 'tracking' trajectory is generated from the target path, with the compensation parameters undisclosed and no baseline or ablation. The real-vehicle experiments are the most solid part of the paper but are orthogonal to the proposed mechanism, since the lead-lag criterion was applied only in post-processing. I recommend rejection, but I note that the error-analysis dataset and the dot-product criterion could form the basis of a more modest paper if the authors (i) disclose and tune all parameters, (ii) add a realistic vehicle model or closed-loop field tests with the compensation active, and (iii) include a with/without compensation comparison. The heavy self-citation of the authors' own arXiv preprints (refs. [30]-[33]) and the promotional framing around the 'preemptive holistic collaborative system' should also be tightened in any resubmission."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Short version: the paper has a genuine empirical core—real-vehicle experiments on straight, curved, and combined tracks with lead/lag error analyzed as a post-processing step—but the central claim, that the proposed lead-lag judgment plus real-time acceleration compensation is validated, is not supported. The 1.21 m simulation result is essentially a self-demonstration.\n\nWhat is new: not much in principle. The lead-lag criterion is the projection of the position difference onto the path tangent, i.e., along-track error, and the compensation law in Eq. (7) is a deadband proportional controller with an undefined time constant T and no disclosed gains. That is standard longitudinal control. The paper does not compare against any baseline or prior spatiotemporal tracking method. What it does well: the field experiments are described concretely, with three speeds, three repeated runs, and separate error components. The lead/lag error time series are useful for understanding how temporal error accumulates in a real vehicle. That part is honest and reproducible in spirit. The citation pattern is fine; self-citations to the group's prior preemptive-system work are plausible context.\n\nWhere it falls: the real-vehicle experiments only used LQR+dual PID; the proposed compensation was not active, so those experiments cannot validate it. The Section 4 simulation, which is the only test of the mechanism, says the vehicle speed and heading angle were adaptively adjusted according to path curvature. That is a curvature-following generator, not a vehicle model with actuator lag, tire dynamics, or lateral-longitudinal coupling. No ablation (compensation on/off), no parameter values, no time step, no baseline. So the reported 1.21 m error could come from the trajectory-generation rule, not the compensation. Conclusion item 2's phrase 'validating its effectiveness' is too strong.\n\nThe stress-test note is right on target. The reader's concerns are fair. I do not see circularity in the real-vehicle data, but the simulation is partly circular because the same law it claims to test generates the trajectory.\n\nWho it is for: someone working on cooperative ramp merging who wants empirical error patterns from a simple LQR+PID vehicle. Not for someone looking for a validated control contribution. With a proper ablation and a realistic vehicle simulator, the central idea might be worth a short paper; as it stands, it needs major revision or rejection.\n\nMy recommendation: send it to peer review if you have reviewers who will demand an ablation and a real vehicle model; expect heavy revision. For the current version, a desk reject is defensible.","headline":"Real-vehicle error data is the one solid piece, but the lead-lag compensation is never actually tested, and the 1.21 m simulation result is not evidence.","tokens_in":10846,"tokens_out":3901,"would_cite":false,"duration_ms":40410,"reading_group":"no","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"A lead-lag judgment mechanism plus real-time acceleration compensation keeps a vehicle's spatiotemporal tracking error to about 1.21 meters in a 100 km/h highway simulation, after real-vehicle tests showed the error grows with speed and…","keywords":["spatiotemporal trajectory tracking","lead-lag judgment","acceleration compensation","LQR lateral control","dual PID longitudinal control","highway ramp merging","connected autonomous vehicles","trajectory tracking error"],"falsifier":"Run the same 100 km/h highway simulation with a validated vehicle dynamics model (e.g., a bicycle model with steering actuator delay, tire slip, and longitudinal-lateral coupling) and measure the maximum lead/lag error; if it substantially exceeds 1.21 m or the acceleration compensation induces oscillation, the paper's central effectiveness claim is not transferable to real vehicles.","tokens_in":9909,"feed_emoji":"🚗","tokens_out":6213,"duration_ms":239124,"temperature":0.7,"pith_summary":"This paper claims that trajectory tracking for autonomous vehicles should control not just where a vehicle is but when it gets there, and offers a concrete way to do so. Its lead-lag judgment mechanism compares the vehicle's position with the target trajectory point at the same time, decides whether the vehicle is ahead or behind using a dot product with the trajectory's tangent, and applies real-time acceleration compensation only when the longitudinal offset exceeds half a meter. Real-vehicle experiments on straight, curved, and combined tracks show that spatiotemporal errors grow with speed and alignment complexity, reaching about two meters in combined straight-curve segments. In a simulation on real highway alignment data at a desired 100 km/h, the mechanism holds the maximum spatiotemporal error to approximately 1.21 meters. If these numbers transfer to real vehicles, the method gives cooperative ramp merging a quantitative spatiotemporal margin to plan against.","feed_headline":"Lead-lag fix holds trajectory error near 1.21 meters at 100 km/h","feed_subtitle":"A dot-product judge asks whether a vehicle is ahead or behind its target point, then adjusts acceleration to close the gap in time.","key_machinery":"The load-bearing mechanism is the lead-lag judgment criterion built on tangent and normal projections of the position-difference vector. With $U_\\tau$ the unit tangent to the target trajectory and $U_n$ its perpendicular, the signs of $\\overrightarrow{PP'}\\cdot U_\\tau$ and $\\overrightarrow{PP'}\\cdot U_n$ classify the tracked point as lead/lag and left/right relative to the target point at the same time. The longitudinal projection feeds a dead-zone acceleration compensator: no correction within $\\pm 0.5$ m, and a quadratic compensation term $\\ddot{a}_{\\mathrm{Compensation}}(t)$ beyond that, so the vehicle's progress is pulled back toward the target trajectory's time schedule. This turns spatiotemporal tracking into a feedback problem on a scalar time-offset variable.","core_discovery":"The central discovery is that a temporal error signal—how far ahead or behind the vehicle is along the target trajectory—can be extracted from the same dot-product geometry used for lateral error and then closed-loop corrected with a simple acceleration law. At each time, the vector from target point to tracked point is projected onto the unit tangent of the target trajectory; a positive projection means the vehicle is ahead (lead), a negative means behind (lag), and the projection magnitude is the longitudinal spatiotemporal error. A perpendicular projection gives left/right lateral error. When the longitudinal error exceeds ±0.5 m, a quadratic acceleration compensation term is added: negative compensation for lead, positive for lag. The paper validates this with low-speed real-vehicle tests using LQR lateral control and dual PID longitudinal control, and with a 100 km/h simulation on a real highway alignment with 2% random speed noise, reporting a maximum spatiotemporal error of about 1.21 m.","pith_inferences":["If the simulated 1.21 m error transfers to practice, it corresponds to roughly 44 ms of travel time at 100 km/h; merging controllers would still need to add communication and actuation latency margins on top, a point the paper leaves implicit.","The dot-product lead-lag criterion is sensor-agnostic and not tied to highway geometry; the same projection could serve as a lightweight temporal-error estimator for drones, mobile robots, or any trajectory follower with an onboard position estimate.","The simulation's idealized vehicle response—speed and heading adjusted only by curvature plus 2% speed noise—omits actuator lag and tire dynamics; a hardware-in-the-loop test with a validated dynamics model would show how much of the 1.21 m survives.","The acceleration compensation formula contains a tuning parameter $\\omega$ that the paper does not systematically sweep; relating it to vehicle responsiveness and jerk limits could turn the mechanism into a design rule rather than a tuned controller."],"forward_implications":["At a desired highway speed of 100 km/h on a real road alignment, with a 2% random speed error, the maximum spatiotemporal (lead/lag) error reported is about 1.21 m, bounded by the threshold plus accumulated correction error.","The paper's real-vehicle experiments show that spatiotemporal error increases with speed and alignment complexity, peaking around 2 m on combined straight-curve segments, so timing control matters most precisely where merging is hardest.","The ±0.5 m dead-zone means small deviations are ignored, which the authors argue prevents overcorrection and instability during normal driving.","Because the mechanism outputs a scalar longitudinal time-offset, it gives a directly usable spatiotemporal distance for cooperative ramp merging in the preemptive holistic collaborative system.","The decoupled control architecture—LQR for lateral, dual PID for longitudinal, plus temporal compensation—is shown in real-vehicle tests to keep lateral and heading errors small, so the temporal layer can be added to existing spatial trackers."],"supporting_citations":[{"why":"Defines the preemptive holistic collaborative system whose correct operation requires spatiotemporal trajectory tracking.","marker":"[30]"},{"why":"Provides the holistic real-time data-sharing view that motivates including the time dimension in trajectory tracking.","marker":"[31]"},{"why":"Establishes preemptive conflict resolution as the control setting where temporal deviations matter for safety.","marker":"[32]"},{"why":"Predefines cooperative highway ramp merging trajectories, the application that gives the spatiotemporal error a safety meaning.","marker":"[33]"}],"fun_headline_variants":["Dot-product judge closes lead-lag gap to 1.21 m","Spatiotemporal tracking gets real-time lead-lag correction","Acceleration tweak fixes vehicle lead/lag on ramps","Highway merge tracking error cut by dot-product method"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing assumption is the simulation's vehicle model, where speed and heading angle respond to path curvature and a small random speed error without real actuator lag, tire slip, or coupling between lateral and longitudinal control; if actual vehicle dynamics differ from that idealization, the measured 1.21 m error may not appear in practice.","fun_headline_variants_meta":{"raw":{"variants":["Dot-product judge closes lead-lag gap to 1.21 m","Spatiotemporal tracking gets real-time lead-lag correction","Acceleration tweak fixes vehicle lead/lag on ramps","Highway merge tracking error cut by dot-product method"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000255,"raw_usage":{"total_tokens":1601,"prompt_tokens":1005,"completion_tokens":596,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":621,"completion_tokens_details":{"reasoning_tokens":525}},"tokens_in":621,"tokens_out":596,"duration_ms":6328,"temperature":1.0,"reasoning_tokens":525,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-08T23:57:27.953424+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Run the same 100 km/h highway simulation with a validated vehicle dynamics model (e.g., a bicycle model with steering actuator delay, tire slip, and longitudinal-lateral coupling) and measure the maximum lead/lag error; if it substantially exceeds 1.21 m or the acceleration compensation induces oscillation, the paper's central effectiveness claim is not transferable to real vehicles.","supporting_citations":[{"cited_title":"Preemptive Holistic Collaborative System and Its Application in Road Transportation","cited_arxiv_id":null,"evidence_quote":"Defines the preemptive holistic collaborative system whose correct operation requires spatiotemporal trajectory tracking."},{"cited_title":"Holistic view of the road transportation system based on real -time data sharing mechanism","cited_arxiv_id":null,"evidence_quote":"Provides the holistic real-time data-sharing view that motivates including the time dimension in trajectory tracking."},{"cited_title":"Preemptive Conflict Resolution in Road Transport: A Holistic Approach for Enhancing Efficiency and Safety [M]","cited_arxiv_id":null,"evidence_quote":"Establishes preemptive conflict resolution as the control setting where temporal deviations matter for safety."},{"cited_title":"Optimizing Highway Ramp Merge Safety and Efficiency via Spatiotemporal Cooperative Control and Vehicle -Road Coordination [J]","cited_arxiv_id":null,"evidence_quote":"Predefines cooperative highway ramp merging trajectories, the application that gives the spatiotemporal error a safety meaning."}],"review_version":1}