Pith. sign in

REVIEW 5 major objections 5 minor 3 cited by

A risk-aware cost term inside QP-MPC yields safer, smoother autonomous-driving trajectories in simulation.

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 ↗

Integrates a dynamic hazard field into QP-MPC trajectory planning and reports smoother, safer lane changes, overtaking, and intersection crossings in simulation.

T0 review reviewed 2026-08-05 challenge →

load-bearing objection The integration of a dynamic hazard field into QP-MPC is a plausible incremental idea, but the paper as written is unverifiable because the central equations are missing and the benchmark comparison is uncontrolled. the 5 major comments →

arxiv 2509.00643 v1 pith:SNLMLYJM submitted 2025-08-31 cs.RO cs.SYeess.SY

A Risk-aware Spatial-temporal Trajectory Planning Framework for Autonomous Vehicles Using QP-MPC and Dynamic Hazard Fields

Zhen Tian , Zhihao Lin , Dezong Zhao , Christos Anagnostopoulos , Qiyuan Wang , Wenjing Zhao , Xiaodan Wang , Chongfeng Wei This is my paper
classification cs.RO cs.SYeess.SY
keywords autonomous drivingtrajectory planningmodel predictive controlquadratic programmingdynamic hazard fieldspace-time graphcollision avoidancedriving comfort
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 establish that a dynamic hazard field can be embedded directly into a quadratic-programming model predictive controller's cost function, so that the optimizer itself minimizes collision risk rather than relying on a separate safe-path filter. The proposed pipeline scores spatial risk with a dynamic hazard field, filters candidate quintic-polynomial trajectories through a space-time graph, and then optimizes a weighted objective covering safety, efficiency, comfort, and duration. In simulations of lane changing, overtaking, and intersection crossing, the authors report that this formulation stays within curvature limits, produces stable velocity and jerk profiles, and avoids predicted collision regions where several generic optimization baselines collide or oscillate. If correct, the framework offers a way to make risk awareness a smooth, differentiable part of trajectory optimization without extra sampling or nonlinear programming.

Core claim

The central claim is that a dynamic hazard field can be expressed as a time-integrated cost term and embedded directly into the QP-MPC objective function. The DHF assigns a scalar risk to every position in Frenet coordinates by summing static hazards from road boundaries and fixed obstacles with dynamic hazards from predicted moving-vehicle positions. This risk term sits alongside smoothness, efficiency, comfort, and duration costs, so the optimizer minimizes hazard exposure instead of merely tracking a precomputed safe path. In the paper's simulations of lane changing, overtaking, and intersection crossing, this formulation keeps velocity, acceleration, and jerk profiles stable, keeps curva

What carries the argument

The dynamic hazard field H(l,d,τ) is the load-bearing object: a scalar risk map over lateral offset l, longitudinal position d, and prediction time τ that sums a static hazard from road boundaries and immobile obstacles with a dynamic hazard from predicted moving-vehicle positions whose spread depends on relative velocity. It enters the planner through the QP-MPC objective, a weighted sum of sub-costs for smoothness (jerk), hazard exposure, efficiency (speed tracking), comfort (acceleration), duration, and tracking error, constrained by speed, acceleration, jerk, curvature, and maximum hazard limits. Because the hazard term is a smooth cost rather than an external constraint, the optimizer c

Load-bearing premise

The entire claimed advantage rests on the hand-tuned formulas and weights in the hazard field and objective function; if those numbers do not transfer to real traffic, the safety and comfort improvements may vanish outside the simulated scenarios.

What would settle it

Ablate the dynamic hazard term by setting its weight to zero (or setting the dynamic hazard to zero) in the QP-MPC objective and re-run the same lane-change, overtaking, and intersection scenarios. If the DHF term contributes, the ablated planner should show measurably higher collision risk, more overlaps with predicted collision regions, or higher jerk; if the trajectories are nearly identical, the paper's central claim that the DHF cost delivers the risk awareness is not supported.

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

If this is right

  • Risk awareness can be treated as a differentiable cost term inside a QP, so safer trajectories are obtained at quadratic-programming speeds instead of through expensive sampling or nonlinear optimization.
  • The DHF is expressed in path-relative coordinates, so the same hazard-cost formulation can be reused in any Frenet-based planner that has access to predicted obstacle states.
  • In the tested scenarios, the proposed planner keeps curvature below limits and avoids predicted collision regions where several baselines cross them, meaning the framework could reduce the need for emergency corrections downstream.
  • The benchmark results suggest that generic optimizers without a risk-shaped cost either collide, exceed curvature limits, or oscillate in these scenarios, while the risk-aware objective produces smoother velocity and jerk profiles.

Where Pith is reading between the lines

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

  • Inference: because the hazard field is a smooth function of predicted obstacle positions, the same cost term could be plugged into other gradient-based planners such as nonlinear MPC or iterative LQR; comparing solve times and trajectory quality across optimizers would test this directly.
  • Inference: the reported gains depend on hand-set hazard parameters and objective weights, so a natural stress test is to vary those parameters or run the planner on recorded real-traffic data to see whether the safety and comfort margins persist outside the tuned simulation settings.
  • Inference: the obstacle prediction model used for the ST graph is deliberately simple (piecewise constant acceleration plus a random term); replacing it with a learned distributional predictor would likely reduce conservative buffering and improve performance in dense intersections, though the paper does not test this.
  • Inference: the lane-change-time results hint at a tunable speed-versus-comfort trade-off—the fastest baseline achieves shorter times at the cost of high jerk—so exposing the objective weights as a user setting could let the same framework switch between sporty and comfortable driving.
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

5 major / 5 minor

Summary. The paper proposes a risk-aware spatial-temporal trajectory planning framework for autonomous vehicles. The framework combines a dynamic hazard field (DHF) for spatial risk assessment, an ST graph for temporal safety checking, quintic polynomial sampling for candidate generation, and a QP-MPC formulation that integrates a weighted sum of safety, efficiency, comfort, and timing sub-costs. The authors claim, based on MATLAB simulations in lane-changing, overtaking, and intersection-crossing scenarios, that the proposed method outperforms six benchmark optimizers (DE, PSO, IP, GA, PS, AS) in efficiency, stability, and comfort. The central contribution is the embedding of a DHF-based risk cost directly into the QP-MPC objective.

Significance. If the proposed DHF-enhanced QP-MPC framework were fully specified and the comparative evaluation were properly controlled, the work would offer a useful integration of risk-field awareness with real-time quadratic programming, and the multi-scenario validation would be a strength. The paper does have a clear architecture, a sensible decomposition into spatial and temporal planning, and an explicit attempt to account for risk in the optimization cost. However, the manuscript currently omits several load-bearing equations and does not specify the benchmark optimization setups, so the claimed empirical superiority is not yet substantiated. Reproducibility and verifiability are the main weaknesses; the ideas are potentially valuable but require substantial technical clarification.

major comments (5)
  1. [Section III, Eq. (3)] The static and dynamic hazard fields are announced as H(l,d,τ) = Hs(l,d) + Hd(l,d,τ), but the actual expressions for Hs and Hd are not displayed. The text jumps directly to 'where (lm, dm) ... Cm ... ηl,m ...' without presenting the functional forms. Since the DHF is the core novelty, the constraint H(l,d,t) ≤ Hmax in Eq. (16), and the hazard sub-cost Jh are all undefined without these formulas. Please provide complete equations and define every symbol (Cm, ηl,m, ηd,m, Dn, ζn, λn, Ln, β).
  2. [Section V, Eq. (16)] The objective J(G) = wsJs + whJh + weJe + wcJc + wtJt + wtret is introduced, but the following sentence 'The individual sub-cost functions are defined as:' is not followed by any equations. The definitions of Js, Jh, Je, Jc, Jt, and et are missing, and no numerical values for the weights are given. This makes it impossible to verify the claimed balance of safety, efficiency, and comfort, or to reproduce the optimization. Please add the explicit sub-cost formulas and the selected weight values, and clarify how these terms are encoded in the QP matrices Q and c in Eqs. (20)-(23).
  3. [Section VI, Table I and Figs. 7-12] The comparative evaluation is not controlled. The reported metrics—longitudinal/lateral acceleration, jerk, and lane-change time—are exactly the quantities penalized in Eq. (16), yet the paper never states what objective each baseline (DE, PSO, IP, GA, PS, AS) was asked to minimize, nor whether the same constraints (curvature limits, Hmax, speed/acceleration bounds, ST safe corridor) were imposed. If the baselines optimize different objectives, the results in Table I may simply reflect objective alignment rather than superior risk-aware planning. Please specify the full optimization setup for each baseline and, ideally, include a variant of the proposed method without the DHF term to isolate its contribution.
  4. [Table I vs. Fig. 10(d)] There is an internal inconsistency in the overtaking results. Table I reports 'IP' with a dash for lane-change time and notes 'a collision occurred', but the text accompanying Fig. 10(d) states that 'The SQP, IP, and PS stay close to zero' and describes the IP as stable and comfortable. This contradiction directly affects the collision-avoidance claim. Please correct the table or the figure discussion and explain whether IP actually collided in this scenario.
  5. [Section IV, Eqs. (5)-(6)] The obstacle prediction model is incomplete. The text says 'Each driving style is predicted using a simple assumed piecewise function:' followed by 'where si(t) is ... and frandom predicts future positions...', but the actual equation is not shown. Consequently the ST obstacle region Φi(t) in Eq. (7) and the safe corridor in Eq. (10) rest on an undefined prediction model. Please provide the explicit piecewise formula, the definition of frandom, and the uncertainty buffer ϵ(t), since temporal safety depends on these quantities.
minor comments (5)
  1. [Throughout] The manuscript contains numerous typos and grammatical issues, e.g., 'comforts' for comfort, 'safty' in Section IV, 'interSection' in Section VI, 'addressesing' in the Conclusion, and 'HDF' for 'DHF' in Section II. A careful proofread is needed.
  2. [Eq. (17)] The kinematic model in Eq. (17) is incomplete. It includes the term 'tan(δk)∆t,' but the update equation for θk is not written, and L (wheelbase) is unused. Please provide the full discrete kinematic equations.
  3. [Section I / Section VII] The paper structure in Section I says 'Section VII describes QP-based Model Predictive control', but Section VII is the Conclusion. Please correct the cross-reference.
  4. [Section V] The abbreviation 'SQP' is used for the proposed 'spatial-temporal-based QP-MPC'. This conflicts with standard usage of SQP for Sequential Quadratic Programming and may confuse readers. Please choose a less ambiguous acronym.
  5. [Figs. 7-12] Several figure captions are incomplete or inconsistent (e.g., Fig. 7 caption says 'HV', while the text refers to the AV; Fig. 12 lists '(e)' for lateral jerk but the subfigure is labeled '(f)' in the text). Please harmonize captions and in-text references.

Circularity Check

1 steps flagged

Claimed comfort/efficiency outperformance is partly self-validating: evaluation metrics are the sub-costs the proposed QP-MPC objective minimizes, and baselines' objectives are unspecified.

specific steps
  1. self definitional [Section V, Eq. (16) and sub-cost definitions; Section VI-B, Table I]
    "min J(G) = wsJs + whJh + weJe + wcJc + wtJt + wtret ... where Js reflects smoothness through jerk minimization. ... Je measures efficiency with a target speed vtarget and weighting λ. Jc computes passenger comfort through minimized accelerations. ... Table I compares average longitudinal acceleration (xa), lateral acceleration (ya), longitudinal jerk (xj), lateral jerk (yj), and lane change time (T)."

    The proposed method's objective explicitly minimizes jerk (Js), acceleration-based comfort (Jc), and efficiency/target-speed (Je), and the paper's headline claim of 'outperforming ... in terms of efficiency, stability, and comfort' is supported almost entirely by metrics that are exactly those quantities (average acceleration, average jerk, lane-change time). The benchmark algorithms (DE, PSO, IP, GA, PS, AS) are generic optimizers and the paper never states what objective each baseline minimized or whether the same constraints were imposed. Under those conditions the reported advantage is a by-construction consequence of optimizing the evaluation metrics, not an independent prediction of superior planning.

full rationale

The paper's spatial-temporal framework (DHF + ST graph + QP-MPC) is self-contained and does not rely on self-citation for its core mechanism. The DHF cost and constraints are stated in Eqs. (3), (16), and the ST-graph corridor in Eqs. (6)-(10); no 'uniqueness theorem' or prior authors' ansatz is load-bearing. The only significant circularity is in the comparative evaluation: the comfort/stability/efficiency metrics (jerk, accelerations, lane-change time) are the same sub-objectives being minimized in Eq. (16). Because the baseline optimizers' objectives are not specified, the comparison does not demonstrate that the proposed risk-aware cost is superior; it demonstrates that minimizing a cost tends to yield low values of that cost. Safety and curvature feasibility are validated by collision-free results and curvature plots, which are independent of this circularity, so the paper is not wholly circular. Score 6 reflects partial reduction of the central outperformance claim to the optimized objective.

Axiom & Free-Parameter Ledger

6 free parameters · 5 axioms · 1 invented entities

The central claim relies on hand-chosen weights, hazard scaling parameters, simplified obstacle prediction, and a linearized kinematic model, none of which are given concrete values or independently validated. No code or data is shipped, so the free-parameter counts are lower bounds on what is needed to reproduce the results.

free parameters (6)
  • Objective weights ws, wh, we, wc, wt, wtr = not reported
    Eq. (16) weights balance smoothness, hazard, efficiency, comfort, and time; no values or tuning procedure are given, yet they determine the reported trade-offs.
  • Hazard field scaling parameters C_m, eta_l,m, eta_d,m, D_n, zeta_n, lambda_n, beta = not reported
    Section III defines the DHF in terms of these parameters but does not state how they are set or validated.
  • Sampling bounds l_min, l_max, d_min, d_max, tau_min, tau_max = not reported
    Section IV defines the sampling domain for candidate trajectories; bounds are not given, affecting the trajectory set.
  • MPC weighting matrices Q, R, Qf = not reported
    Section V uses Q, R, and Qf as quadratic penalties; values are omitted.
  • Target speed v_target and efficiency weight lambda = not reported
    Used in the efficiency sub-cost J_e; no values given.
  • Obstacle uncertainty buffer epsilon(t) = not reported
    Eq. (6) adds a time-dependent buffer to predicted obstacle positions; its shape is not specified.
axioms (5)
  • standard math Frenet frame and cubic spline reference path transformations
    Section III uses Eq. (1)-(2) to derive curvature and coordinate conversion; standard differential geometry.
  • standard math Quintic polynomial can represent feasible trajectories satisfying boundary conditions
    Section IV Eq. (11)-(14); standard interpolation result.
  • domain assumption Obstacle future motion follows a simple piecewise constant-acceleration model with a random term
    Section IV Eq. (5) predicts positions from initial velocity and acceleration plus f_random; this is a major simplification of real driving behavior.
  • domain assumption Vehicle motion is captured by a discrete kinematic bicycle model linearized about the current state
    Section V Eq. (17)-(18) ignores tire dynamics, slip, and model mismatch.
  • ad hoc to paper The dynamic hazard field values and weights are meaningful measures of real-world risk and comfort
    Sections III and V define H and the objective; no independent validation establishes that the chosen formula correlates with actual collision risk or passenger comfort.
invented entities (1)
  • Dynamic Hazard Field (DHF) no independent evidence
    purpose: A scalar risk field used in the cost function to penalize proximity to static and dynamic obstacles during QP-MPC optimization.
    It is a mathematical construct introduced by the paper for the planner's objective; no physical measurement or external falsifiable prediction is attached to it.

reviewed 2026-08-05 · how reviews work

0 comments
Cite this review

Pith. "Pith review of A Risk-aware Spatial-temporal Trajectory Planning Framework for Autonomous Vehicles Using QP-MPC and Dynamic Hazard Fields." pith.science (2026). https://pith.science/paper/SNLMLYJM

@misc{pith2026250900643,
  author       = {Pith},
  title        = {Pith review of: A Risk-aware Spatial-temporal Trajectory Planning Framework for Autonomous Vehicles Using QP-MPC and Dynamic Hazard Fields},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/SNLMLYJM}},
  note         = {Machine review of arXiv:2509.00643}
}
Share X Bluesky LinkedIn Reddit HN
read the original abstract

Trajectory planning is a critical component in ensuring the safety, stability, and efficiency of autonomous vehicles. While existing trajectory planning methods have achieved progress, they often suffer from high computational costs, unstable performance in dynamic environments, and limited validation across diverse scenarios. To overcome these challenges, we propose an enhanced QP-MPC-based framework that incorporates three key innovations: (i) a novel cost function designed with a dynamic hazard field, which explicitly balances safety, efficiency, and comfort; (ii) seamless integration of this cost function into the QP-MPC formulation, enabling direct optimization of desired driving behaviors; and (iii) extensive validation of the proposed framework across complex tasks. The spatial safe planning is guided by a dynamic hazard field (DHF) for risk assessment, while temporal safe planning is based on a space-time graph. Besides, the quintic polynomial sampling and sub-reward of comforts are used to ensure comforts during lane-changing. The sub-reward of efficiency is used to maintain driving efficiency. Finally, the proposed DHF-enhanced objective function integrates multiple objectives, providing a proper optimization tasks for QP-MPC. Extensive simulations demonstrate that the proposed framework outperforms benchmark optimization methods in terms of efficiency, stability, and comfort across a variety of scenarios likes lane-changing, overtaking, and crossing intersections.

discussion (0)

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

Forward citations

Cited by 3 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score.

  1. Attention and Risk-Aware Decision Framework for Safe Autonomous Driving

    cs.RO 2025-09 reject novelty 4.0

    An improved PPO for highway driving uses a hybrid risk field, attention modules, a balanced reward, and a rule-based safety filter to reduce collisions and speed up training in simulation.

  2. Adaptive Evolution Factor Risk Ellipse Framework for Reliable and Safe Autonomous Driving

    cs.RO 2025-09 reject novelty 4.0

    An adaptive risk-field-plus-MPC controller with a sigmoid evolution factor and TTC/TWH-based risk ellipses is claimed to achieve collision-free overtaking and lane changes in simulation.

  3. Scenario-based Decision-making Using Game Theory for Interactive Autonomous Driving: A Survey

    cs.RO 2025-09 reject novelty 1.0

    A scenario-based survey of game-theoretic autonomous driving decision-making that claims comprehensiveness but is undermined by a non-systematic methodology and numerous internal errors.

Reference graph

Works this paper leans on

29 extracted references · 25 canonical work pages · cited by 3 Pith papers

  1. [1]

    Bio-inspired hybrid path planning for efficient and smooth robotic navigation: F. yuan et al

    F. Yuan et al., “Bio-inspired hybrid path planning for efficient and smooth robotic navigation: F. yuan et al.” International Journal of Intelligent Robotics and Applications, pp. 1–31, 20 25

    work page

  2. [2]

    Trajectory planning for an autonomous vehicle in spatially constrained environments,

    Y. Guo, D. Yao et al., “Trajectory planning for an autonomous vehicle in spatially constrained environments,” IEEE Trans. Intell. Transp. Syst.s, vol. 23, no. 10, pp. 18 326–18 336, 2022

    work page 2022

  3. [3]

    Dpl-slam: enhancing dy- namic point-line slam through dense semantic methods,

    Z. Lin, Q. Zhang, Z. Tian, P. Yu, and J. Lan, “Dpl-slam: enhancing dy- namic point-line slam through dense semantic methods,” IEEE Sensors Journal, vol. 24, no. 9, pp. 14 596–14 607, 2024

    2024

  4. [4]

    Enhanced visual slam for collision-free driving with lightweight autonomous cars,

    Z. Lin, Z. Tian, Q. Zhang, H. Zhuang, and J. Lan, “Enhanced visual slam for collision-free driving with lightweight autonomous cars,” Sensors, vol. 24, no. 19, p. 6258, 2024

    2024

  5. [5]

    Slam2: Simultaneous localization and multimode mapping for indoor dynamic environments,

    Z. Lin, Q. Zhang, Z. Tian, P. Yu, Z. Ye, H. Zhuang, and J. Lan, “Slam2: Simultaneous localization and multimode mapping for indoor dynamic environments,” Pattern Recognition, vol. 158, p. 111054, 2025. [6 ] Y. Liu, B. Zhou et al. , “ Dynamic lane- changing trajectory planning for autonomous vehicles based on discrete global trajectory,” IEEE Trans. Intell...

    work page 2025

  6. [7]

    Efficient and safe planner for automated driving on ramps considering unsatisfication,

    Q. Li et al., “Efficient and safe planner for automated driving on ramps considering unsatisfication,” arXiv preprint arXiv:2504.15320, 2025

    Pith/arXiv arXiv 2025

  7. [8]

    Autonomous vehicle cut-in algorithm for lane- merging scenarios via policy-based reinforcement learning nested within finite-state machine,

    S. Hwang, K. Lee et al., “Autonomous vehicle cut-in algorithm for lane- merging scenarios via policy-based reinforcement learning nested within finite-state machine,” IEEE Trans. Intell. Transp. Syst.s, vol. 23, no. 10, pp. 17 594–17 606, 2022

    work page 2022

  8. [9]

    A graph-search based trajectory planning approach with longitudianl priority,

    R. Guo and S. Yu, “A graph-search based trajectory planning approach with longitudianl priority,” in Proceeding of the CAA International Conference on Vehicular Control and Intelligence, 2022, pp. 1–6

    work page 2022

  9. [10]

    Adaptive field effect planner for safe interactive au- tonomous driving on curved roads,

    Q. Li et al., “Adaptive field effect planner for safe interactive au- tonomous driving on curved roads,” arXiv preprint arXiv:2504.14747 , 2025

    Pith/arXiv arXiv 2025

  10. [11]

    Multi-agent monte carlo tree search for safe decision making at unsignalized intersections,

    Z. Lin et al., “Multi-agent monte carlo tree search for safe decision making at unsignalized intersections,” 2025

    work page 2025

  11. [12]

    Safety- critical multi-agent mcts for mixed traffic coordination at unsignalized intersections,

    Z. Lin, J. Lan, C. Anagnostopoulos, Z. Tian, and D. Flynn, “Safety- critical multi-agent mcts for mixed traffic coordination at unsignalized intersections,” IEEE Transactions on Intelligent Transportation Systems, 2025

    work page 2025

  12. [13]

    Receding horizon control using graph search for multi-agent trajectory planning,

    P. Scheffe, M. V. A. Pedrosa et al., “Receding horizon control using graph search for multi-agent trajectory planning,” IEEE Trans. Control Syst. Technol., vol. 31, no. 3, pp. 1092–1105, 2023

    work page 2023

  13. [14]

    A novel direct trajectory planning approach based on generative adversarial networks and rapidly-exploring random tree,

    C. Zhao, Y. Zhu et al., “A novel direct trajectory planning approach based on generative adversarial networks and rapidly-exploring random tree,” IEEE Trans. Intell. Transp. Syst.s, vol. 23, no. 10, pp. 17 910– 17 921, 2022

    work page 2022

  14. [15]

    Human-machine cooperative trajectory planning and tracking for safe automated driving,

    C. Huang, H. Huang et al., “Human-machine cooperative trajectory planning and tracking for safe automated driving,” IEEE Trans. Intell. Transp. Syst.s, vol. 23, no. 8, pp. 12 050–12 063, 2022. [1 6 ] T. Ma and F. Song, “ A trajectory privacy protection method based on random sampling differential privacy,” ISPRS International Journal of Geo-Information, v...

    work page 2022

  15. [18]

    Trajectory and velocity planning method of emergency rescue vehicle based on segmented three-dimensional quartic bezier curve,

    T. Chen, Y. Cai et al., “Trajectory and velocity planning method of emergency rescue vehicle based on segmented three-dimensional quartic bezier curve,” IEEE Trans. Intell. Transp. Syst.s, vol. 24, no. 3, pp. 3461– 3475, 2023

    work page 2023

  16. [19]

    Cooperative driving of automated vehicles using b-splines for trajectory planning,

    R. van Hoek, J. Ploeg et al., “ Cooperative driving of automated vehicles using b-splines for trajectory planning,” IEEE Trans. Intell. Veh., vol. 6, no. 3, pp. 594–604, 2021. [2 0] T. Lee and S. B. Choi, “ Real- time optimization of gear shift trajectories using quadratic programming for electric vehicles with dual clutch transmission,” IEEE Trans. Veh. ...

    work page 2021

  17. [21]

    A two-stage miqp-based optimization approach for coordinating automated electric vehicles in confined sites,

    S. Kojchev, R. Hult et al., “A two-stage miqp-based optimization approach for coordinating automated electric vehicles in confined sites,” IEEE Trans. Intell. Transp. Syst.s, vol. 25, no. 2, pp. 2061–2075, 2024

    work page 2061

  18. [22]

    Path-speed decoupling planning method based on risk cooperative game for intelligent vehicles,

    Z. Zhang, C. Wang et al., “Path-speed decoupling planning method based on risk cooperative game for intelligent vehicles,” IEEE Transac- tions on Transportation Electrification, 2023

    work page 2023

  19. [23]

    Decoupled real-time trajectory planning for multiple autonomous mining trucks in unloading areas,

    Q. Yang, Y. Ai et al., “Decoupled real-time trajectory planning for multiple autonomous mining trucks in unloading areas,” IEEE Trans. Intell. Veh., 2023

    work page 2023

  20. [24]

    An improved model predictive control- based trajectory planning method for automated driving vehicles under 12 uncertainty environments,

    T. Qie, W. Wang et al., “An improved model predictive control- based trajectory planning method for automated driving vehicles under 12 uncertainty environments,” IEEE Trans. Intell. Transp. Syst.s, vol. 24, no. 4, pp. 3999–4015, 2023

    work page 2023

  21. [25]

    A mean-field-game-integrated mpc-qp framework for collision-free multi-vehicle control,

    L. Zheng, X. Wang, F. Li, Z. Mao, Z. Tian, Y. Peng, F. Yuan, and C. Yuan, “A mean-field-game-integrated mpc-qp framework for collision-free multi-vehicle control,” Drones, vol. 9, no. 5, p. 375, 2025

    work page 2025

  22. [26]

    A hybrid trajectory planning strategy for intelligent vehicles in on-road dynamic scenarios,

    M. Wang, L. Zhang et al., “ A hybrid trajectory planning strategy for intelligent vehicles in on-road dynamic scenarios,” IEEE Trans. Veh. Technol., vol. 72, no. 3, pp. 2832–2847, 2022. [2 7 ] Z. Zhang, L. Zhang et al. , “ An enabling trajectory planning scheme for lane change collision avoidance on highways,” IEEE Trans. Intell. Veh., vol. 8, no. 1, pp. ...

    work page 2022

  23. [29]

    Traffic signal optimization based on fuzzy control and differential evolution algorithm,

    H. Lin, Y. Han et al., “Traffic signal optimization based on fuzzy control and differential evolution algorithm,” IEEE Trans. Intell. Transp. Syst.s, vol. 24, no. 8, pp. 8555–8566, 2023

    work page 2023

  24. [30]

    Adaptivity-enhanced path tracking system for autonomous vehicles at high speeds,

    G. Huang, X. Yuan et al., “Adaptivity-enhanced path tracking system for autonomous vehicles at high speeds,” IEEE Trans. Intell. Veh., vol. 5, no. 4, pp. 626–634, 2020

    work page 2020

  25. [31]

    Forces nlp: An efficient implementation of interior-point methods for multistage nonlinear nonconvex programs,

    A. Zanelli, A. Domahidi et al., “Forces nlp: An efficient implementation of interior-point methods for multistage nonlinear nonconvex programs,” International Journal of Control, vol. 93, no. 1, pp. 13–29, 2020

    work page 2020

  26. [32]

    Optimal trajectory planning and robust tracking using vehicle model inversion,

    S. Victor, J.-B. Receveur et al., “Optimal trajectory planning and robust tracking using vehicle model inversion,” IEEE Trans. Intell. Transp. Syst.s, vol. 23, no. 5, pp. 4556–4569, 2022

    work page 2022

  27. [33]

    Optimal control by pattern search optimization method,

    K. Kandhway, “Optimal control by pattern search optimization method,” in Proceeding of the International Conference on Computational Intel- ligence and Communication Networks, 2022, pp. 224–228

    work page 2022

  28. [34]

    Active set strategy for high- dimensional non-convex sparse optimization problems,

    A. Boisbunon, R. Flamary et al., “Active set strategy for high- dimensional non-convex sparse optimization problems,” in Proceeding of the International Conference on Acoustics, Speech and Signal Pro- cessing, 2014, pp. 1517–1521. Zhen Tian received his bachelor degree in electronic and electrical engineering from the University of Strathclyde,Glasgow, U....

    work page 2014

  29. [2020]

    degree with the School of Engineering , Cardiff Univer- sity, Cardiff, UK

    She is currently pursuing the Ph.D. degree with the School of Engineering , Cardiff Univer- sity, Cardiff, UK. Her research interests focus on human-robot collaboration, task planning and human factors. Chongfeng Wei received his Ph.D. degree in me- chanical engineering from the University of Birm- ingham in 20 1 5 . He is now a Senior Lecturer ( As- soci...

    work page

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