Pith. sign in

REVIEW 4 major objections 6 minor 61 references

This paper embeds a time-to-collision penalty inside a double-quintic lane-change solver, claiming it eliminates sub-second TTC violations while keeping trajectories smooth and comfortable.

Reviewed by Pith at T0; open to challenge. T0 means a machine referee read the full paper against a public rubric. the ladder, T0–T4 →

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

An optimization-based double quintic lane-change planner that penalizes low time-to-collision is proposed and tested in simulation.

T0 review reviewed 2026-08-05 challenge →

load-bearing objection A real but incremental contribution—TTC-in-the-loop double-quintic planning—undermined by self-referential simulation; deserves a serious referee, not a desk reject. the 4 major comments →

arxiv 2509.00582 v1 pith:QH4CRCTA submitted 2025-08-30 cs.RO cs.SYeess.SY

Safe and Efficient Lane-Changing for Autonomous Vehicles: An Improved Double Quintic Polynomial Approach with Time-to-Collision Evaluation

Rui Bai , Rui Xu , Teng Rui , Jiale Liu , Qi Wei Oung , Hoi Leong Lee , Zhen Tian , Fujiang Yuan This is my paper
classification cs.RO cs.SYeess.SY
keywords autonomous drivinglane changingdouble quintic polynomialtime-to-collisiontrajectory optimizationmixed trafficjerk minimizationsafety planning
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

The paper tries to show that lane-changing safety does not have to be a post-hoc check: a time-to-collision penalty can be baked directly into the closed-form double-quintic polynomial solver that generates the trajectory. It claims this is the first time an analytic TTC term is embedded in the polynomial coefficient optimization, producing C2-continuous paths that keep TTC above a user-set threshold in all 45 simulated scenarios. The contribution matters because it promises real-time, safety-aware trajectory generation without the replanning loops or black-box behavior of learning-based planners. If true, it would give autonomous vehicles a lightweight, interpretable way to balance safety, efficiency, and passenger comfort during lane changes in mixed traffic.

Core claim

The core claim is that replacing the direct boundary-condition inversion a = M⁻¹b with a constrained optimization that minimizes a combined smoothness-plus-TTC cost yields trajectories that are proactively safe rather than merely smooth. The TTC penalty is a piecewise-quadratic function φᵢ(t) = [max{0, T_safe − TTCᵢ(t)}]², which is C¹ and differentiable with respect to the polynomial coefficients, so gradient-based solvers can steer coefficients away from low-TTC regions. The paper reports that this eliminates sub-second TTC violations in all 45 simulated scenarios, more than doubles the minimum obstacle gap from 8.75 m to 19.53 m, cuts the share of low-TTC (<3 s) time steps from 64% to 42%,

What carries the argument

The central object is the double-quintic polynomial trajectory, split into two quintic segments at a switching time so the maneuver has explicit phase control. The key mechanism is the analytic TTC penalty embedded in the coefficient optimization: a differentiable term added to the jerk-smoothness cost, with coefficients initialized from the closed-form boundary inversion and refined by an interior-point solver. This converts safety from a post-hoc filter into a continuous gradient signal that shapes the trajectory during generation.

Load-bearing premise

The surrounding vehicles are partly generated by applying a time delay and lateral offset to the ego's own planned trajectory (Eq. 44), so avoiding them is easier than avoiding truly independent human-driven vehicles; if real HDVs do not follow the ego's path, the claimed TTC guarantees may not transfer.

What would settle it

Run the same planner in closed loop against pre-recorded NGSIM trajectories or an independent car-following model that does not depend on the ego vehicle's planned path, then count sub-second TTC violations and minimum gaps. If any sub-second TTC violation appears or the minimum gap drops below 10 m in those scenarios, the central safety claim is refuted.

Watch this falsifier. Get emailed when new claim-graph text bears on it.

If this is right

  • Lane-change planners could enforce user-defined safety margins directly in the trajectory-generation step, removing the need for frequent reactive replanning.
  • The TTC threshold T_safe becomes a tunable dial: raising it produces more conservative, smoother paths, while lowering it yields faster, more aggressive maneuvers, as shown in the ablation study.
  • The same double-quintic formulation generalizes from a single lane change to overtaking, modeled as two consecutive lane changes, and to partial or half-lane maneuvers.
  • The optimization converges reliably within roughly 20 iterations, suggesting the method is light enough for embedded or real-time deployment on resource-constrained hardware.
  • Because the penalty is differentiable and smooth, the approach can be grafted onto other polynomial or spline trajectory families, not just quintics.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • The paper's safety guarantees are demonstrated in simulations where surrounding HDV trajectories are partly generated by time-shifting and laterally offsetting the ego vehicle's own planned trajectory (Eq. 44); this makes avoidance easier than it would be with fully independent human drivers, and transferring the guarantees to real traffic would require testing against independent, pre-recorded or
  • The TTC penalty as defined rewards keeping TTC above one global threshold; in real traffic, a more informative safety measure might combine TTC with lateral spacing, road curvature, and prediction uncertainty, which the current analytic form does not capture.
  • The claim of being 'the first' to embed an analytic TTC penalty in a double-quintic solver is a literature claim, not a mathematical one; a careful reader should check whether earlier polynomial-planner papers used similar differentiable safety penalties before treating that novelty as established.
  • The differentiable-penalty trick could plausibly be extended to other safety metrics, such as predicted minimum distance or jerk-weighted collision risk, opening a family of 'safety-in-the-cost' planners.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit.

Referee Report

4 major / 6 minor

Summary. The paper proposes an improved double-quintic polynomial lane-change planner in which a time-to-collision (TTC) penalty is embedded directly into the coefficient optimization. The central claims are that this produces C2-continuous trajectories, eliminates sub-second TTC violations, keeps TTC above a user-defined threshold in all 45 simulated scenarios, and is the first analytic TTC penalty inside a closed-form double-quintic solver. The method is compared against quintic, Bezier, and B-spline baselines in lane-changing and overtaking scenarios, with additional experiments on ramps, an unsignalized intersection, an ablation study, and a highway-env/NGSIM extension.

Significance. If the central claim were established, embedding TTC directly into polynomial coefficient optimization would be a useful, low-cost alternative to post-hoc safety validation. The authors are transparent about several limitations, notably the simplified constant-speed/no-sideslip vehicle model and simulation-only evaluation, and the later NGSIM replay and highway-env experiments are steps toward more realistic interactions. However, the load-bearing evaluation is circular and the safety guarantees are not supported by the evidence. The main contribution as stated, namely a safety-guaranteed planner that completely eliminates low-TTC events, does not follow from the presented method or experiments.

major comments (4)
  1. [Section 5, Eq. (44)] The HDV trajectories are generated as r_HDV(t) = r_ego(t+tau) + delta, i.e., every obstacle is a time-delayed, laterally shifted copy of the ego vehicle's own planned trajectory. The 45-scenario claim that TTC remains above threshold is therefore a property of the optimizer's own output, not of interaction with independent traffic. The NGSIM replay in Section 6 is described as replacing synthetic HDV generators only 'in some simulations,' so it does not repair the core benchmark. This makes the headline safety result untestable and not transferable to real mixed traffic.
  2. [Section 4.4 / Section 6.3] The TTC term is a soft penalty, not a hard constraint: Eq. (22) is stated as an inequality but is not enforced as such; the cost is the piecewise-quadratic penalty in Eqs. (23)-(24). The paper's own ablation reports a minimum gap of about 1 m in the Fast-Lane-Change overtaking configuration and gaps below 4 m for the Aggressive-TTC configuration, directly contradicting the abstract's 'completely eliminate sub-second TTC violations' and the introduction's claim of a 'minimum gap to every HDV above 10m.' At best, the method reduces low-TTC exposure; it does not guarantee safety.
  3. [Section 4.4, Eqs. (21)-(26)] The exact TTC formula and penalty are not fully present in the manuscript. Eq. (21) uses d_i(t) and Delta v_i(t) without defining them; Eqs. (23) and (24) are blank in the text; Eq. (26) is incomplete. Since every claimed result depends on this cost, the derivation is not reproducible from the manuscript and the 'analytic TTC penalty' central to the contribution cannot be examined.
  4. [Section 4.2 / Eq. (19) / Section 7.1] The planner assumes constant longitudinal velocity, x(t)=v t, and neglects sideslip. However, the experiments show longitudinal speed variation (e.g., 22-25 m/s in Figs. 25 and 27), braking HDVs, and speed-oscillating vehicles. The authors themselves state in Section 7.1 that the simplified model does not capture real-world steering, acceleration, and tire-road constraints. The safety and feasibility claims therefore go beyond what the stated model can support. The internal inconsistency between the constant-speed planning model and the time-varying speed in the evaluation should be resolved.
minor comments (6)
  1. [Section 4.4.1 vs. 4.7.1] The two theoretical subsections contradict each other: Section 4.4.1/4.4.2 says the problem is not globally convex, while Section 4.7.1 states the overall cost 'remains convex with respect to the polynomial coefficients.' Also, the proof of Theorem 1 claims the smoothness term is quadratic and 'hence coercive,' which is not guaranteed for a positive semi-definite quadratic form on an affine subspace.
  2. [Section 5, Eq. (42)] Road boundaries are defined as offsets of the ego trajectory, r_left(t)=r_ego(t)+w n(t). This makes the road geometry depend on the planned path itself; in real applications lane boundaries are independent of the ego vehicle. This should be clarified or replaced with a fixed road model.
  3. [Figures 25-27] The labels Vx and Vy appear inconsistent with Eq. (3): the text calls Vx 'lateral speed' and Vy 'longitudinal speed' in places, while Eq. (3) and the vehicle dynamics use x as longitudinal and y as lateral. The captions also refer to 'lateral acceleration alpha' and 'lateral velocity Vy' interchangeably. Please unify notation.
  4. [Table 6] The column header 'Improvement' shows -9.9% for minimum TTC and minimum distance, which is a degradation, not an improvement. The text explains this as 'decisive' behavior, but the table as presented is misleading. Also, the method name 'stable TTC' is not introduced elsewhere.
  5. [Section 1, Contribution 3] The paper states 'We release an open MATLAB testbed' but no repository link or availability statement is provided. This should be added if the claim is to hold.
  6. [Throughout] There are duplicated sections (Section 2.5 and Section 4.6 both titled 'Clarification of the methodological novelty'), duplicated existence/convergence discussions (Sections 4.4.1-4.4.3 and 4.7.1-4.7.4), and several typos ('effieicncy', 'comforts', 'stabilization' for 'robustness'). A thorough editing pass is needed.

Circularity Check

2 steps flagged

Safety benchmark is self-referential: HDVs and road boundaries are generated from the ego trajectory itself.

specific steps
  1. self definitional [Section 5, Eq. (44)]
    "HDV trajectories are generated by applying a time delay τ and a lateral offset δ to the trajectory of the AV: rHDV(t) = rego(t + τ) + δ. (44)"

    The obstacle field is constructed from the ego vehicle's own planned trajectory. The TTC to each 'HDV' is therefore d_i(t)=||r_ego(t)-r_ego(t+τ)-δ||, a function of the ego trajectory alone. The planner can shape r_ego so that this autocorrelation-based TTC stays above any threshold; there are no independent HDV dynamics (no braking, acceleration, or cut-ins). The reported claim that the method 'completely eliminates sub-second TTC violations and keep TTC above the user-defined threshold in all 45 simulated scenarios' is thus a property of the optimizer's own output, not a prediction about mixed traffic. The validation is self-referential and cannot transfer to real traffic.

  2. self definitional [Section 5, Eq. (42)]
    "The road boundaries are generated by offsetting the trajectory of the AV along its unit normal: rleft(t) = rego(t) + w n(t), rright(t) = rego(t) - w n(t), (42)"

    If lane boundaries are defined as lateral offsets of the planned AV trajectory, then every planned ego trajectory is inside the boundaries by construction. Statements that the 'ego vehicle consistently follows the planned trajectory across all ramp conditions without deviating beyond the safe lane boundaries' are tautological and provide no independent safety evidence. This is another instance of the evaluation being defined in terms of the planner's own output.

full rationale

The core double-quintic solver and the TTC penalty are internally coherent as an optimization formulation; optimizing coefficients against a TTC-inclusive cost is not itself circular. The circularity lies in the empirical validation. Eq. (44) defines every HDV as a time-delayed, laterally offset copy of the ego trajectory, so the obstacle field is a deterministic function of the planner's own output. The headline all-45-scenario safety result therefore reduces to an autocorrelation property of the optimized ego path, not to interaction with independent traffic. Eq. (42) similarly defines road boundaries from the ego trajectory, making 'staying within boundaries' tautological. The paper notes NGSIM replay is used only 'in some simulations,' leaving the primary 45-scenario benchmark on the self-referential synthetic generator. The ablation study also reports minimum gaps near 1 m in aggressive/fast configurations, further undermining the 'safety-guaranteed' interpretation even within the same setup. No load-bearing self-citation chain was found; the problem is the self-referential evaluation, not citation practice.

Axiom & Free-Parameter Ledger

5 free parameters · 4 axioms · 0 invented entities

The central claim rests on manually chosen cost weights and thresholds, a kinematic model that ignores longitudinal acceleration and sideslip, and a synthetic obstacle model that reuses the ego trajectory. These are not independently validated.

free parameters (5)
  • smoothness weight λ1 = 1.0 (in Sec. 6.6 experiment)
    Chosen by hand; no tuning rule given.
  • TTC penalty weight λ2 = 5.0 (Sec. 6.6); recommended range [0.1,10]
    Manual selection; affects whether TTC violations are avoided.
  • safety threshold Tsafe = 3.0 s baseline; 1.5 / 4.5 s in ablation
    User-defined; the claimed guarantee depends on this value.
  • safety distance (ablation) = 5.0 / 8.0 m
    Used in ablation but not clearly part of the cost function.
  • switching / phase times T1,T2,T3,T4 = chosen per scenario (e.g., 2.0-3.4 s)
    Hand-picked; affects feasibility and comfort.
axioms (4)
  • domain assumption Constant longitudinal velocity and no lateral slip for the ego vehicle
    Invoked in Sec. 4.2 and used throughout the polynomial and TTC derivations; violates real vehicle dynamics.
  • ad hoc to paper HDV trajectories are generated as r_HDV(t) = r_ego(t+τ) + δ
    Sec. 5, Eq. 44. Makes obstacles depend on the planned ego trajectory, biasing the safety evaluation.
  • ad hoc to paper Road boundaries are offsets of the AV trajectory
    Sec. 5, Eq. 42. As written, this makes lane boundaries depend on the planned path; likely a modeling error.
  • standard math LICQ and second-order sufficiency hold at local minima for sensitivity analysis
    Sec. 4.4.3 invokes standard regularity assumptions without verifying them for this problem.

reviewed 2026-08-05 · how reviews work

0 comments
Cite this review

Pith. "Pith review of Safe and Efficient Lane-Changing for Autonomous Vehicles: An Improved Double Quintic Polynomial Approach with Time-to-Collision Evaluation." pith.science (2026). https://pith.science/paper/QH4CRCTA

@misc{pith2026250900582,
  author       = {Pith},
  title        = {Pith review of: Safe and Efficient Lane-Changing for Autonomous Vehicles: An Improved Double Quintic Polynomial Approach with Time-to-Collision Evaluation},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/QH4CRCTA}},
  note         = {Machine review of arXiv:2509.00582}
}
Share X Bluesky LinkedIn Reddit HN
read the original abstract

Autonomous driving technology has made significant advancements in recent years, yet challenges remain in ensuring safe and comfortable interactions with human-driven vehicles (HDVs), particularly during lane-changing maneuvers. This paper proposes an improved double quintic polynomial approach for safe and efficient lane-changing in mixed traffic environments. The proposed method integrates a time-to-collision (TTC) based evaluation mechanism directly into the trajectory optimization process, ensuring that the ego vehicle proactively maintains a safe gap from surrounding HDVs throughout the maneuver. The framework comprises state estimation for both the autonomous vehicle (AV) and HDVs, trajectory generation using double quintic polynomials, real-time TTC computation, and adaptive trajectory evaluation. To the best of our knowledge, this is the first work to embed an analytic TTC penalty directly into the closed-form double-quintic polynomial solver, enabling real-time safety-aware trajectory generation without post-hoc validation. Extensive simulations conducted under diverse traffic scenarios demonstrate the safety, efficiency, and comfort of the proposed approach compared to conventional methods such as quintic polynomials, Bezier curves, and B-splines. The results highlight that the improved method not only avoids collisions but also ensures smooth transitions and adaptive decision-making in dynamic environments. This work bridges the gap between model-based and adaptive trajectory planning approaches, offering a stable solution for real-world autonomous driving applications.

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

61 extracted references · 46 canonical work pages · 1 internal anchor

  1. [1]

    Transportation Research Part A: Policy and Practice 199, 104525 (2025)

    Hao, H., Yao, E., Pan, L., Chen, R., Wang, Y., Xiao, H.: Exploring heterogeneous drivers and barriers in maas bundle subscriptions based on the willingness to shift to maas in one-trip scenarios. Transportation Research Part A: Policy and Practice 199, 104525 (2025)

    work page 2025

  2. [2]

    arXiv preprint arXiv:2505.17685 (2025)

    Zeng, S., Chang, X., Xie, M., Liu, X., Bai, Y., Pan, Z., Xu, M., Wei, X.: Future- sightdrive: Thinking visually with spatio-temporal cot for autonomous driving. arXiv preprint arXiv:2505.17685 (2025)

    Pith/arXiv arXiv 2025

  3. [3]

    arXiv preprint arXiv:2409.05352 (2024)

    Zeng, S., Chang, X., Liu, X., Pan, Z., Wei, X.: Driving with prior maps: Unified vector prior encoding for autonomous vehicle mapping. arXiv preprint arXiv:2409.05352 (2024)

    arXiv 2024

  4. [4]

    Meiring, G.A.M., Myburgh, H.C.: A review of intelligent driving style analysis systemsandrelatedartificialintelligencealgorithms.Sensors 15(12),30653–30682 (2015)

    work page 2015

  5. [5]

    IEEE Transactions on Intelligent Transportation Systems 23(7), 6011–6026 (2021)

    Ning, H., Yin, R., Ullah, A., Shi, F.: A survey on hybrid human-artificial intelli- gence for autonomous driving. IEEE Transactions on Intelligent Transportation Systems 23(7), 6011–6026 (2021)

    work page 2021

  6. [6]

    In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pp

    Zhu, S., Wang, G., Blum, H., Liu, J., Song, L., Pollefeys, M., Wang, H.: Sni-slam: Semantic neural implicit slam. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pp. 21167–21177 (2024)

    work page 2024

  7. [7]

    Pattern Recognition 158, 111054 (2025)

    Lin, Z., Zhang, Q., Tian, Z., Yu, P., Ye, Z., Zhuang, H., Lan, J.: Slam2: Simul- taneous localization and multimode mapping for indoor dynamic environments. Pattern Recognition 158, 111054 (2025)

    work page 2025

  8. [8]

    arXiv preprint arXiv:2504.03724 (2025)

    Wang, Z., Feng, P., Lin, Y., Cai, S., Bian, Z., Yan, J., Zhu, X.: Crowdvlm-r1: Expanding r1 ability to vision language model for crowd counting using fuzzy group relative policy reward. arXiv preprint arXiv:2504.03724 (2025)

    arXiv 2025

  9. [9]

    Information Sciences 662, 120286 (2024)

    Zhou, Y., Xia, H., Yu, D., Cheng, J., Li, J.: Outlier detection method based on high-density iteration. Information Sciences 662, 120286 (2024)

    work page 2024

  10. [10]

    Computational Visual Media 10(3), 593–608 (2024)

    Zhang, F., Chen, G., Wang, H., Zhang, C.: Cf-dan: Facial-expression recognition based on cross-fusion dual-attention network. Computational Visual Media 10(3), 593–608 (2024)

    work page 2024

  11. [12]

    Biomimetics 65 10(5) (2025) https://doi.org/10.3390/biomimetics10050342

    Zhang, K., Yuan, F., Jiang, Y., Mao, Z., Zuo, Z., Peng, Y.: A particle swarm optimization-guided ivy algorithm for global optimization problems. Biomimetics 65 10(5) (2025) https://doi.org/10.3390/biomimetics10050342

    work page doi:10.3390/biomimetics10050342 2025

  12. [13]

    Liang,B., Yuan,F.,Deng, J.,Wu, Q., Gao, J.: Cs-pbft: a comprehensive scoring- based practical byzantine fault tolerance consensus algorithm. J. Supercomput. 81, 859 (2025)

    work page 2025

  13. [14]

    Computers 14(7) (2025) https://doi.org/10.3390/ computers14070256

    Yuan, F., Huang, X., Jiang, H., Jiang, Y., Zuo, Z., Wang, L., Wang, Y., Gu, S., Peng, Y.: An xlstm–xgboost ensemble model for forecasting non-stationary and highly volatile gasoline price. Computers 14(7) (2025) https://doi.org/10.3390/ computers14070256

    work page 2025

  14. [15]

    arXiv preprint arXiv:2505.19522 (2025)

    Hu, J., Zeng, H., Tian, Z.: Applications and effect evaluation of generative adversarial networks insemi-supervised learning. arXiv preprint arXiv:2505.19522 (2025)

    Pith/arXiv arXiv 2025

  15. [16]

    IEEE Sensors Journal 24(9), 14596– 14607 (2024)

    Lin, Z., Zhang, Q., Tian, Z., Yu, P., Lan, J.: Dpl-slam: enhancing dynamic point- line slam through dense semantic methods. IEEE Sensors Journal 24(9), 14596– 14607 (2024)

    work page 2024

  16. [17]

    arXiv preprint arXiv:2504.15320 (2025)

    Li, Q., Tian, Z., Wang, X., Yang, J., Lin, Z.: Efficient and safe planner for automated driving on ramps considering unsatisfication. arXiv preprint arXiv:2504.15320 (2025)

    Pith/arXiv arXiv 2025

  17. [18]

    arXiv preprint arXiv:2504.14747 (2025)

    Li, Q., Tian, Z., Wang, X., Yang, J., Lin, Z.: Adaptive field effect planner for safe interactive autonomous driving on curved roads. arXiv preprint arXiv:2504.14747 (2025)

    Pith/arXiv arXiv 2025

  18. [19]

    In: 2025 4th International Conference on New Energy System and Power Engi- neering (NESP), pp

    Liu, Y., Tian, Z., Yang, J., Lin, Z.: Data-driven evolutionary game-based model predictive control for hybrid renewable energy dispatch in autonomous ships. In: 2025 4th International Conference on New Energy System and Power Engi- neering (NESP), pp. 482–490 (2025). https://doi.org/10.1109/NESP65198.2025. 11041353

    work page arXiv 2025

  19. [20]

    IEEE Robotics and Automation Letters 8(5), 2708–2715 (2023)

    Vishnu, C., Abhinav, V., Roy, D., Mohan, C.K., Babu, C.S.: Improving multi- agent trajectory prediction using traffic states on interactive driving scenarios. IEEE Robotics and Automation Letters 8(5), 2708–2715 (2023)

    work page 2023

  20. [21]

    arXiv preprint arXiv:2501.01886 (2025)

    Tian, Z., Lin, Z., Zhao, D., Zhao, W., Flynn, D., Ansari, S., Wei, C.: Evaluating scenario-based decision-making for interactive autonomous driving using rational criteria: A survey. arXiv preprint arXiv:2501.01886 (2025)

    Pith/arXiv arXiv 2025

  21. [22]

    Engineering Applications of Artificial Intelligence 160, 111874 (2025)

    Jiang, L., Zhang, W., Wang, H., Zhang, F.: Robust memory-based graph neu- ral networks for noisy and sparse graphs. Engineering Applications of Artificial Intelligence 160, 111874 (2025)

    work page 2025

  22. [23]

    IEEE 66 transactions on intelligent transportation systems 23(6), 4909–4926 (2021)

    Kiran, B.R., Sobh, I., Talpaert, V., Mannion, P., Al Sallab, A.A., Yogamani, S., P´erez, P.: Deep reinforcement learning for autonomous driving: A survey. IEEE 66 transactions on intelligent transportation systems 23(6), 4909–4926 (2021)

    work page 2021

  23. [24]

    Engineering 21, 75–91 (2023)

    Wu, J., Huang, Z., Hu, Z., Lv, C.: Toward human-in-the-loop ai: Enhancing deep reinforcement learning via real-time human guidance for autonomous driving. Engineering 21, 75–91 (2023)

    work page 2023

  24. [25]

    Sensors 24(19), 6258 (2024) https://doi.org/10.3390/s24196258

    Lin, Z., Tian, Z., Zhang, Q., Zhuang, H., Lan, J.: Enhanced visual slam for collision-free driving with lightweight autonomous cars. Sensors 24(19), 6258 (2024) https://doi.org/10.3390/s24196258

    work page doi:10.3390/s24196258 2024

  25. [26]

    Drones 9(5), 375 (2025) https://doi.org/10.3390/drones9050375

    Zheng, L., Wang, X., Li, F., Mao, Z., Tian, Z., Peng, Y., Yuan, F., Yuan, C.: A mean-field-game-integrated MPC-QP framework for collision-free multi-vehicle control. Drones 9(5), 375 (2025) https://doi.org/10.3390/drones9050375

    work page doi:10.3390/drones9050375 2025

  26. [27]

    arXiv preprint arXiv:2408.08242 (2024)

    Lin, Z., Tian, Z., Zhang, Q., Ye, Z., Zhuang, H., Lan, J.: A conflicts-free, speed- lossless kan-based reinforcement learning decision system for interactive driving in roundabouts. arXiv preprint arXiv:2408.08242 (2024). Accepted to IEEE Transactions on Intelligent Transportation Systems (2025)

    Pith/arXiv arXiv 2024

  27. [28]

    International Journal of Intelligent Robotics and Applications (2025)https://doi

    Yuan, F., Lin, Z., Tian, Z., Chen, B., Zhou, Q.,Yuan, C., Sun, H., Huang, Z., et al.: Bio-inspired hybrid path planning for efficient and smooth robotic navigation. International Journal of Intelligent Robotics and Applications (2025)https://doi. org/10.1007/s41315-025-00479-7

    work page doi:10.1007/s41315-025-00479-7 2025

  28. [29]

    IEEE Transactions on Intelligent Transportation Systems (2025)

    Lin, Z., Lan, J., et al.: Safety-critical multi-agent mcts for mixed traffic coordination at unsignalized intersections. IEEE Transactions on Intelligent Transportation Systems (2025). Early Access / accepted

    work page 2025

  29. [30]

    In: Proceedings of the AAAI Conference on Artificial Intelligence, vol

    Liu, J., Wang, G., Jiang, C., Liu, Z., Wang, H.: Translo: A window-based masked point transformer framework for large-scale lidar odometry. In: Proceedings of the AAAI Conference on Artificial Intelligence, vol. 37, pp. 1683–1691 (2023)

    work page 2023

  30. [31]

    In: The Thirteenth International Conference on Learning Representations

    Zhou, X., Zhang, M., Lee, Z., Ye, W., Zhang, S.: Hademif: Hallucination detec- tion and mitigation in large language models. In: The Thirteenth International Conference on Learning Representations

    work page

  31. [32]

    In: Proceedings of the IEEE/CVF International Conference on Computer Vision, pp

    Liu, J., Wang, G., Liu, Z., Jiang, C., Pollefeys, M., Wang, H.: Regformer: an efficient projection-aware transformer network for large-scale point cloud regis- tration. In: Proceedings of the IEEE/CVF International Conference on Computer Vision, pp. 8451–8460 (2023)

    work page 2023

  32. [33]

    IEEE Transactions on Circuits and Systems for Video Technology 33(9), 4496–4506 (2023)

    Zhang, F., Chen, G., Wang, H., Li, J., Zhang, C.: Multi-scale video super- resolution transformer with polynomial approximation. IEEE Transactions on Circuits and Systems for Video Technology 33(9), 4496–4506 (2023)

    work page 2023

  33. [34]

    The 67 Visual Computer, 1–12 (2025)

    Jiang, L., Wang, X., Zhang, F., Zhang, C.: Transforming time and space: efficient video super-resolution with hybrid attention and deformable transformers. The 67 Visual Computer, 1–12 (2025)

    work page 2025

  34. [35]

    In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pp

    Liu, J., Wang, G., Ye, W., Jiang, C., Han, J., Liu, Z., Zhang, G., Du, D., Wang, H.: Difflow3d: Toward robust uncertainty-aware scene flow estimation with itera- tive diffusion-based refinement. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pp. 15109–15119 (2024)

    work page 2024

  35. [36]

    EADReg: Probabilistic Correspondence Generation with Efficient Autoregressive Diffusion Model for Outdoor Point Cloud Registration

    Gong, L., Liu, J., Ma, J., Liu, L., Wang, Y., Wang, H.: Eadreg: Probabilistic cor- respondence generation with efficient autoregressive diffusion model for outdoor point cloud registration. arXiv preprint arXiv:2411.15271 (2024)

    work page internal anchor Pith review Pith/arXiv arXiv 2024

  36. [37]

    Pattern Recognition 161, 111312 (2025)

    Lu, W., Wang, J., Wang, T., Zhang, K., Jiang, X., Zhao, H.: Visual style prompt learning using diffusion models for blind face restoration. Pattern Recognition 161, 111312 (2025)

    work page 2025

  37. [38]

    IEEE Transactions on Knowledge and Data Engineering (2025)

    Zhou, X., Ye, W., Lee, Z., Zou, L., Zhang, S.: Valuing training data via causal inference for in-context learning. IEEE Transactions on Knowledge and Data Engineering (2025)

    work page 2025

  38. [39]

    In: Proceedings of the 62nd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), pp

    Zhou, X., Ye, W., Wang, Y., Jiang, C., Lee, Z., Xie, R., Zhang, S.: Enhancing in- context learning via implicit demonstration augmentation. In: Proceedings of the 62nd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), pp. 2810–2828 (2024)

    work page 2024

  39. [40]

    In: Proceedings ofthe Computer Vision and Pattern Recognition Conference, pp

    Liu, J., Han, J., Liu, L., Aviles-Rivero, A.I., Jiang, C., Liu, Z., Wang, H.: Mamba4d: Efficient 4d point cloud video understanding with disentangled spatial- temporal state space models. In: Proceedings ofthe Computer Vision and Pattern Recognition Conference, pp. 17626–17636 (2025)

    work page 2025

  40. [41]

    arXiv preprint arXiv:2403.06467 (2024)

    Liu, J., Yu, R., Wang, Y., Zheng, Y., Deng, T., Ye, W., Wang, H.: Point mamba: A novel point cloud backbone based on state space model with octree-based ordering strategy. arXiv preprint arXiv:2403.06467 (2024)

    Pith/arXiv arXiv 2024

  41. [42]

    arXiv preprint arXiv:2504.00364 (2025)

    Chen, Y.-H., Liu, S., Xiao, W., Belta, C., Otte, M.: Control barrier functions via minkowski operations for safe navigation among polytopic sets. arXiv preprint arXiv:2504.00364 (2025)

    Pith/arXiv arXiv 2025

  42. [43]

    arXiv preprint arXiv:2503.23267 (2025)

    Liu, S., Xiao, W., Belta, C.A.: Ensuring safe and smooth control in safety-critical systems via filtered control barrier functions. arXiv preprint arXiv:2503.23267 (2025)

    arXiv 2025

  43. [44]

    arXiv preprint arXiv:2503.19205 (2025)

    Liu, S., Belta, C.A.: Risk-aware adaptive control barrier functions for safe control of nonlinear systems under stochastic uncertainty. arXiv preprint arXiv:2503.19205 (2025)

    arXiv 2025

  44. [45]

    In: 2023 62nd IEEE Conference on Decision and 68 Control (CDC), pp

    Liu, S., Xiao, W., Belta, C.A.: Auxiliary-variable adaptive control barrier func- tions for safety critical systems. In: 2023 62nd IEEE Conference on Decision and 68 Control (CDC), pp. 8602–8607 (2023). IEEE

    work page 2023

  45. [46]

    arXiv preprint arXiv:2210.04361 (2022)

    Liu, S., Zeng, J., Sreenath, K., Belta, C.A.: Iterative convex optimization for model predictive control with discrete-time high-order control barrier functions. arXiv preprint arXiv:2210.04361 (2022)

    Pith/arXiv arXiv 2022

  46. [47]

    In: European Conference on Computer Vision, pp

    Li, J., Liao, B., Lu, X., Liu, P., Shen, S., Zhou, Y.: Event-aided time-to-collision estimation for autonomous driving. In: European Conference on Computer Vision, pp. 57–73 (2024). Springer

    work page 2024

  47. [48]

    Artificial Life and Robotics 28(1), 158–165 (2023)

    Nie, X., Liang, Y., Ohkura, K.: Autonomous highway driving using reinforcement learning with safety check system based on time-to-collision. Artificial Life and Robotics 28(1), 158–165 (2023)

    work page 2023

  48. [49]

    In: 2023 IEEE Intelligent Vehicles Symposium (IV), pp

    Lin, P., Javanmardi, E., Tao, Y., Chauhan, V., Nakazato, J., Tsukada, M.: Time- to-collision-aware lane-change strategy based on potential field and cubic poly- nomial for autonomous vehicles. In: 2023 IEEE Intelligent Vehicles Symposium (IV), pp. 1–6 (2023). IEEE

    work page 2023

  49. [50]

    Journal of advanced transportation 2023(1), 6926304 (2023)

    Li, Y., Li, L., Ni, D.: Dynamic trajectory planning for automated lane changing using the quintic polynomial curve. Journal of advanced transportation 2023(1), 6926304 (2023)

    work page 2023

  50. [51]

    In: 2022 7th Interna- tional Conference on Intelligent Informatics and Biomedical Science (ICIIBMS), vol

    Liu,Y., Wang,Y., Wang,X., Tao,C.: Amodel forsafelanechangingofconnected vehicles based on quintic polynomial trajectory planning. In: 2022 7th Interna- tional Conference on Intelligent Informatics and Biomedical Science (ICIIBMS), vol. 7, pp. 51–55 (2022). IEEE

    work page 2022

  51. [52]

    Electronics 14(3), 494 (2025)

    Li, H., Pang, H., Xia, H., Huang, Y., Zeng, X.: Research on trajectory planning method based on b´ezier curves for dynamic scenarios. Electronics 14(3), 494 (2025)

    work page 2025

  52. [53]

    Intelligent Automation & Soft Computing 37(1) (2023)

    Wang, X., Qin, X., Zhang, H., Minchala, L.I.: Cognitive granular-based path plan- ning and tracking for intelligent vehicle with multi-segment bezier curve stitching. Intelligent Automation & Soft Computing 37(1) (2023)

    work page 2023

  53. [54]

    World Electric Vehicle Journal 13(12), 233 (2022)

    Wang, P., Yang, J., Zhang, Y., Wang, Q., Sun, B., Guo, D.: Obstacle-avoidance path-planning algorithm for autonomous vehicles based on b-spline algorithm. World Electric Vehicle Journal 13(12), 233 (2022)

    work page 2022

  54. [55]

    Control Engineering Practice 165, 106546 (2025)

    Jin, L., Qiu, N., Tian, D., Zhang, Q., Teng, F., Jin, B., Xiao, F.: Mpc-based path tracking strategy for 4wid&4wis vehicles using b-spline approximation and state-dependent reference. Control Engineering Practice 165, 106546 (2025)

    work page 2025

  55. [56]

    69 International journal of automotive technology 24(2), 459–468 (2023)

    Meng, T., Yang, T., Huang, J., Jin, W., Zhang, W., Jia, Y., Wan, K., Xiao, G., Yang, D., Zhong, Z.: Improved hybrid a-star algorithm for path plan- ning in autonomous parking system based on multi-stage dynamic optimization. 69 International journal of automotive technology 24(2), 459–468 (2023)

    work page 2023

  56. [57]

    In: 2024 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), pp

    Minaˇr´ık, M., Pˇeniˇcka, R., Von´asek, V., Saska, M.: Model predictive path inte- gral control for agile unmanned aerial vehicles. In: 2024 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), pp. 13144–13151 (2024). IEEE

    work page 2024

  57. [58]

    Information 16(4) (2025) https://doi.org/10.3390/ info16040268

    Yuan, F., Huang, X., Zheng, L., Wang, L., Wang, Y., Yan, X., Gu, S., Peng, Y.: The evolution and optimization strategies of a pbft consensus algorithm for consortium blockchains. Information 16(4) (2025) https://doi.org/10.3390/ info16040268

    work page 2025

  58. [59]

    Algorithms 18(5) (2025)https://doi.org/10.3390/a18050263

    Yuan, F., Zuo, Z., Jiang, Y., Shu, W., Tian, Z., Ye, C., Yang, J., Mao, Z., Huang, X., Gu, S., Peng, Y.: Ai-driven optimization ofblockchain scalability, security, and privacy protection. Algorithms 18(5) (2025)https://doi.org/10.3390/a18050263

    work page doi:10.3390/a18050263 2025

  59. [60]

    Transportation Research Part C: Emerging Technologies 19(6), 1243–1262 (2011)

    Punzo, V., Borzacchiello, M.T., Ciuffo, B.: On theassessment ofvehicletrajectory data accuracy and application to the next generation simulation (ngsim) program data. Transportation Research Part C: Emerging Technologies 19(6), 1243–1262 (2011)

    work page 2011

  60. [61]

    IEEE Transactions on Intelligent Transportation Systems 24(12), 14097–14110 (2023) https://doi.org/10.1109/TITS.2023.3298344

    Lu, G., Hua, J., Zhao, H., Liu, M., Xu, J., Zong, F.: Modeling vehicle paths at intersections: A unified approach based on entrance and exit lanes. IEEE Transactions on Intelligent Transportation Systems 24(12), 14097–14110 (2023) https://doi.org/10.1109/TITS.2023.3298344

    arXiv 2023

  61. [62]

    IEEE Transactions on Intelligent Vehicles 6(3), 594–604 (2021) https://doi.org/10.1109/TIV.2021.3072679 70

    Hoek, R., Ploeg, J., Nijmeijer, H.: Cooperative driving of automated vehicles using b-splines for trajectory planning. IEEE Transactions on Intelligent Vehicles 6(3), 594–604 (2021) https://doi.org/10.1109/TIV.2021.3072679 70

    arXiv 2021

This paper was first reviewed by deepseek-v4-flash on August 5, 2026.