pith. sign in

arxiv: 2508.04436 · v2 · submitted 2025-08-06 · 💻 cs.RO · cs.SY· eess.SY

Reliable and Real-Time Highway Trajectory Planning via Hybrid Learning-Optimization Frameworks

Yujia Lu , Chong Wei , Lu Ma , Lounis Adouane This is my paper

Pith reviewed 2026-05-19 00:20 UTC · model grok-4.3

classification 💻 cs.RO cs.SYeess.SY
keywords highway trajectory planninghybrid learning optimizationmixed-integer quadratic programmingautonomous drivingcollision avoidancereal-time planningsafety constraintsHighD dataset
0
0 comments X

The pith

Hybrid framework splits learning for traffic adaptation from optimization for formal safety in highway planning.

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

The paper establishes a hybrid highway trajectory planning method that assigns traffic-adaptive velocity generation to a learning module while routing all collision avoidance and kinematic feasibility checks to a Mixed-Integer Quadratic Program. This division keeps formal safety constraints active at every step even when many vehicles interact. A geometric linearization cuts the number of integer variables so the optimization solves fast enough for real time. Tests on the HighD dataset report success above 97 percent with an average cycle time near 54 milliseconds and trajectories that stay smooth and feasible. Autonomous highway driving needs exactly this combination of adaptability and guaranteed safety because reaction windows are short and errors carry high cost.

Core claim

The H-HTP framework integrates a learning module that produces a traffic-adaptive velocity profile with an MIQP that enforces every safety-critical decision on collision avoidance and kinematic feasibility. The linearization strategy for vehicle geometry reduces integer variables enough to support real-time solution while preserving the formal safety guarantees of the optimization.

What carries the argument

Mixed-Integer Quadratic Program (MIQP) that uses a linearization of vehicle geometry to encode collision avoidance and kinematic constraints.

If this is right

  • Formal safety constraints remain enforced in any multi-vehicle configuration.
  • Trajectories stay smooth, kinematically feasible, and collision-free.
  • Scenario success rate exceeds 97 percent on the HighD dataset.
  • Average planning cycle completes in about 54 milliseconds.

Where Pith is reading between the lines

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

  • The same split between learned adaptation and verified safety could apply to urban or merging scenarios with denser interactions.
  • The approach suggests that hybrid designs may lower reliance on purely rule-based safety layers in autonomous systems.
  • Evaluating the linearization on datasets with higher traffic density or sensor noise would test its robustness limits.
  • Coupling the planner with real-time perception modules could yield closed-loop autonomy that still carries explicit safety certificates.

Load-bearing premise

Linearizing vehicle geometry reduces the integer variables enough to reach real-time speed without weakening the formal safety guarantees.

What would settle it

A recorded multi-vehicle highway scene in which the planner either outputs a colliding trajectory or exceeds real-time computation limits would disprove the central reliability claim.

Figures

Figures reproduced from arXiv: 2508.04436 by Chong Wei, Lounis Adouane, Lu Ma, Yujia Lu.

Figure 1
Figure 1. Figure 1: Overall Framework of the Proposed Method [PITH_FULL_IMAGE:figures/full_fig_p004_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: Discretization of Vehicle Trajectory trajectory, while the second results from improved convexity in the path optimization formulation and the lightweight design of the end-to-end velocity prediction model. IV. TRAJECTORY GENERATION AND VEHICLE KINEMATIC MODEL Typically, an s-l-t coordinate system is established to fa￾cilitate trajectory planning. In this system, the s-l coordinates define a Frenet frame [… view at source ↗
Figure 3
Figure 3. Figure 3: Illustration of Safety Corridor Search linear approximation method was developed to minimize the number of integer variables involved. Wang et al. [9] adopted a big-M formulation that introduced Q × N binary variables, where Q denotes the number of segments used to piecewise￾linearize the nonlinear function. Compared with their method, our approach reduces the number of binary variables to N, denoted by ki… view at source ↗
Figure 4
Figure 4. Figure 4: Geometric Representation and Lateral Vehicle Kinematics [PITH_FULL_IMAGE:figures/full_fig_p006_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: A Linear Approximation to f1(x) on the heading angle φ arises from the fundamental kinematic relationship tan φ = dl ds = l ′ ⇒ ε = f1(l ′ ) = 1 2 WEVp 1 + (l ′) 2, (9) where WEV denotes the EV’s width. This nonlinear formula￾tion would render the path-planning optimization model non￾linear, significantly compromising computational efficiency. A linear approximation method is developed by leveraging the ty… view at source ↗
Figure 6
Figure 6. Figure 6: Vectorized Representation of Trajectories and Map Features [PITH_FULL_IMAGE:figures/full_fig_p008_6.png] view at source ↗
Figure 7
Figure 7. Figure 7: Scenario 1 Performance. (a) Sequential snapshots at 0.5-second intervals; (b) kinematic profile showing longitudinal and lateral displacement, velocity, [PITH_FULL_IMAGE:figures/full_fig_p010_7.png] view at source ↗
Figure 8
Figure 8. Figure 8: Scenario 2 Performance. (a) Sequential snapshots at 0.5-second intervals; (b) kinematic profile showing longitudinal and lateral displacement, velocity, [PITH_FULL_IMAGE:figures/full_fig_p011_8.png] view at source ↗
Figure 9
Figure 9. Figure 9: Scenario 3 Performance. (a) Sequential snapshots at 0.5-second intervals; (b) kinematic profile showing longitudinal and lateral displacement, velocity, [PITH_FULL_IMAGE:figures/full_fig_p012_9.png] view at source ↗
Figure 10
Figure 10. Figure 10: Distribution of Planning Cycle Runtimes. [PITH_FULL_IMAGE:figures/full_fig_p012_10.png] view at source ↗
Figure 12
Figure 12. Figure 12: Velocity Prediction Error Over Iterations [PITH_FULL_IMAGE:figures/full_fig_p013_12.png] view at source ↗
read the original abstract

Autonomous highway driving involves high-speed safety risks due to limited reaction time, where rare but dangerous events may lead to severe consequences. This places stringent requirements on trajectory planning in terms of both reliability and computational efficiency. This paper proposes a hybrid highway trajectory planning (H-HTP) framework that integrates learning-based adaptability with optimization-based formal safety guarantees. The key design principle is a deliberate division of labor: a learning module generates a traffic-adaptive velocity profile, while all safety-critical decisions including collision avoidance and kinematic feasibility are delegated to a Mixed-Integer Quadratic Program (MIQP). This design ensures that formal safety constraints are always enforced, regardless of the complexity of multi-vehicle interactions. A linearization strategy for the vehicle geometry substantially reduces the number of integer variables, enabling real-time optimization without sacrificing formal safety guarantees. Experiments on the HighD dataset demonstrate that H-HTP achieves a scenario success rate above 97% with an average planning-cycle time of approximately 54 ms, reliably producing smooth, kinematically feasible, and collision-free trajectories in safety-critical highway scenarios.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

2 major / 1 minor

Summary. The paper proposes a hybrid highway trajectory planning (H-HTP) framework that integrates a learning-based module to generate traffic-adaptive velocity profiles with a Mixed-Integer Quadratic Program (MIQP) that enforces all safety-critical constraints including collision avoidance and kinematic feasibility. A linearization strategy for vehicle geometry is introduced to reduce the number of integer variables and enable real-time solving. Experiments on the HighD dataset report scenario success rates above 97% and average planning-cycle times of approximately 54 ms, claiming smooth, kinematically feasible, and collision-free trajectories in safety-critical highway scenarios.

Significance. If the linearization preserves formal safety guarantees, the framework offers a promising division of labor between learning adaptability and optimization-based reliability for high-speed autonomous driving. The use of an external dataset and reported real-time performance metrics are positive indicators of practical relevance, though the central safety claim requires rigorous verification to be load-bearing.

major comments (2)
  1. [Abstract] Abstract: The claim that the linearization strategy for vehicle geometry enables real-time optimization 'without sacrificing formal safety guarantees' is load-bearing for the central assertion that formal safety constraints are always enforced. Collision avoidance between oriented rectangles is non-convex; without an explicit proof or verification (e.g., showing the linearized constraints form a sound over-approximation of the original nonlinear distance constraints for all relative headings and positions), feasible MIQP solutions could correspond to actual collisions in the true geometry. This directly affects the reliability guarantee.
  2. [Experiments] Experiments section: The reported scenario success rate above 97% and average planning time of 54 ms are summarized without error bars, ablation studies on the linearization, or explicit verification that the MIQP solutions remain collision-free under the original nonlinear geometry. This makes it difficult to assess whether the empirical results support the formal safety claim across the tested interactions.
minor comments (1)
  1. [Abstract] The abstract and method description would benefit from a brief reference to the specific linearization technique (e.g., outer approximation or fixed-orientation facets) and any associated conservatism bounds.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the detailed and constructive feedback on our manuscript. The comments highlight important aspects of our safety claims and empirical validation that we will strengthen in the revision. Below we respond point-by-point to the major comments.

read point-by-point responses

  1. Referee: [Abstract] Abstract: The claim that the linearization strategy for vehicle geometry enables real-time optimization 'without sacrificing formal safety guarantees' is load-bearing for the central assertion that formal safety constraints are always enforced. Collision avoidance between oriented rectangles is non-convex; without an explicit proof or verification (e.g., showing the linearized constraints form a sound over-approximation of the original nonlinear distance constraints for all relative headings and positions), feasible MIQP solutions could correspond to actual collisions in the true geometry. This directly affects the reliability guarantee.

    Authors: We agree that an explicit verification of the linearization's soundness is necessary to support the formal safety claim. The linearization in the manuscript approximates oriented vehicle rectangles via conservative axis-aligned bounds in a locally rotated frame, which over-approximates the minimum separation distance. To address the referee's concern, we will add a dedicated subsection (or appendix) in the revised manuscript that formally proves the linearized constraints constitute a sound over-approximation: any point satisfying the MIQP constraints satisfies the original nonlinear collision-avoidance inequalities for all relative headings and positions encountered in highway driving. We will also include a brief geometric argument showing that the approximation error is bounded and does not permit actual collisions. revision: yes

  2. Referee: [Experiments] Experiments section: The reported scenario success rate above 97% and average planning time of 54 ms are summarized without error bars, ablation studies on the linearization, or explicit verification that the MIQP solutions remain collision-free under the original nonlinear geometry. This makes it difficult to assess whether the empirical results support the formal safety claim across the tested interactions.

    Authors: We acknowledge that the current experimental presentation lacks statistical detail and direct verification of the original geometry. In the revised version we will (i) report success rates and planning times with standard deviations or error bars computed over multiple dataset folds or repeated trials, (ii) add an ablation study that isolates the linearization's impact on both runtime and success rate, and (iii) include a post-processing verification step that checks every accepted MIQP solution against the original nonlinear distance constraints, confirming zero violations on the HighD test scenarios. These additions will make the empirical support for the safety claim more transparent. revision: yes

Circularity Check

0 steps flagged

No circularity: hybrid framework with independent dataset evaluation

full rationale

The paper's core derivation separates a learning module for velocity profiles from an MIQP that encodes collision avoidance and kinematic constraints. Safety guarantees are asserted via the MIQP formulation itself rather than being derived from fitted outputs or self-referential definitions. Reported metrics (97% success rate, 54 ms cycle time) are obtained from experiments on the external HighD dataset and are not algebraically forced by any internal parameters or prior self-citations. The linearization step is presented as an engineering choice to reduce integer variables; it does not redefine the safety constraints in terms of the reported performance numbers. No load-bearing equation or claim reduces to a tautology or to quantities fitted inside the same manuscript.

Axiom & Free-Parameter Ledger

0 free parameters · 1 axioms · 0 invented entities

The abstract provides no explicit free parameters, invented entities, or non-standard axioms; the framework rests on standard assumptions of mixed-integer quadratic programming and vehicle kinematic models.

axioms (1)
  • domain assumption Vehicle kinematics and collision avoidance can be expressed as linear or quadratic constraints inside an MIQP
    This assumption underpins the delegation of all safety-critical decisions to the optimizer.

pith-pipeline@v0.9.0 · 5729 in / 1247 out tokens · 54982 ms · 2026-05-19T00:20:54.579477+00:00 · methodology

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

37 extracted references · 37 canonical work pages · 1 internal anchor

  1. [1]

    A review of motion planning for highway autonomous driving,

    L. Claussmann, M. Revilloud, and D. Gruyer, “A review of motion planning for highway autonomous driving,” IEEE Transactions on Intelligent Transportation Systems, vol. 21, no. 5, pp. 1826–1848, 2020

    work page 2020

  2. [2]

    Exact obstacle avoidance for autonomous vehicles in polygonal domains,

    J. Fan, N. Murgovski, and J. Liang, “Exact obstacle avoidance for autonomous vehicles in polygonal domains,” IEEE Transactions on Systems, Man, and Cybernetics: Systems , vol. 54, no. 10, pp. 5964– 5976, 2024

    work page 2024

  3. [3]

    Optimization-based collision avoidance,

    X. Zhang, A. Liniger, and F. Borrelli, “Optimization-based collision avoidance,” IEEE Transactions on Control Systems Technology, vol. 29, no. 3, pp. 972–983, 2021

    work page 2021

  4. [4]

    Efficient optimization-based trajectory planning for unmanned systems in confined environments,

    J. Fan, N. Murgovski, and J. Liang, “Efficient optimization-based trajectory planning for unmanned systems in confined environments,” IEEE Transactions on Intelligent Transportation Systems, vol. 25, no. 11, pp. 18 547–18 560, 2024

    work page 2024

  5. [5]

    Fast collision-free multi-vehicle lane change motion planning and control framework in uncertain environ- ments,

    T. Liu, R. Chai, and S. Chai, “Fast collision-free multi-vehicle lane change motion planning and control framework in uncertain environ- ments,” IEEE Transactions on Industrial Electronics , vol. 71, no. 12, pp. 16 602–16 613, 2024

    work page 2024

  6. [6]

    A successive linearization in feasible set algorithm for vehicle motion planning in unstructured and low-speed scenarios,

    C. Sun, Q. Li, and B. Li, “A successive linearization in feasible set algorithm for vehicle motion planning in unstructured and low-speed scenarios,” IEEE Transactions on Intelligent Transportation Systems , vol. 23, no. 4, pp. 3724–3736, 2022

    work page 2022

  7. [7]

    A nonlinear programing model for collision-free lane-change trajectory planning based on vehicle-to- vehicle communication,

    C. Wei, Y . Wang, and Y . Asakura, “A nonlinear programing model for collision-free lane-change trajectory planning based on vehicle-to- vehicle communication,” Journal of Transportation Safety & Security , vol. 13, no. 9, pp. 936–956, 2019

    work page 2019

  8. [8]

    Vectornet: Encoding hd maps and agent dynamics from vectorized rep- resentation,

    J. Gao, C. Sun, H. Zhao, Y . Shen, D. Anguelov, C. Li, and C. Schmid, “Vectornet: Encoding hd maps and agent dynamics from vectorized rep- resentation,” in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) , June 2020

    work page 2020

  9. [9]

    A convex trajectory planning method for autonomous vehicles considering kinematic feasibility and bi-state obstacles avoidance effectiveness,

    Y . Wang, C. Wei, and S. Li, “A convex trajectory planning method for autonomous vehicles considering kinematic feasibility and bi-state obstacles avoidance effectiveness,” IEEE Transactions on Vehicular Technology, vol. 73, no. 7, pp. 9575–9590, Jul. 2024

    work page 2024

  10. [10]

    Hybrid trajectory planning for autonomous driving in on-road dynamic scenarios,

    W. Lim, S. Lee, and M. Sunwoo, “Hybrid trajectory planning for autonomous driving in on-road dynamic scenarios,” IEEE Transactions on Intelligent Transportation Systems, vol. 22, no. 1, pp. 341–355, 2021

    work page 2021

  11. [11]

    An enabling trajectory planning scheme for lane change collision avoidance on highways,

    Z. Zhang, L. Zhang, and J. Deng, “An enabling trajectory planning scheme for lane change collision avoidance on highways,” IEEE Trans- actions on Intelligent Vehicles , vol. 8, no. 1, pp. 147–158, 2023

    work page 2023

  12. [12]

    A cooperative trajectory planning system based on the passengers’ individual preferences of aggressiveness,

    Y . Yan, J. Wang, and Y . Wang, “A cooperative trajectory planning system based on the passengers’ individual preferences of aggressiveness,”IEEE Transactions on Vehicular Technology, vol. 72, no. 1, pp. 395–406, 2023

    work page 2023

  13. [13]

    Integrated decision making and planning based on feasible region construction for autonomous vehicles considering prediction uncertainty,

    L. Xiong, Y . Zhang, and Y . Liu, “Integrated decision making and planning based on feasible region construction for autonomous vehicles considering prediction uncertainty,” IEEE Transactions on Intelligent Vehicles, vol. 8, no. 11, pp. 4515–4523, 2023

    work page 2023

  14. [14]

    Autonomous vehicles lane-changing tra- jectory planning based on hierarchical decoupling,

    X. Lin, T. Wang, and S. Zeng, “Autonomous vehicles lane-changing tra- jectory planning based on hierarchical decoupling,” IEEE Transactions on Intelligent Transportation Systems, vol. 25, no. 12, pp. 20 741–20 752, 2024

    work page 2024

  15. [15]

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

    Z. Zhang, C. Wang, and W. Zhao, “Path–speed decoupling planning method based on risk cooperative game for intelligent vehicles,” IEEE Transactions on Transportation Electrification, vol. 10, no. 2, pp. 3792– 3806, Jun. 2024

    work page 2024

  16. [16]

    Dynamic lane-changing trajectory planning for autonomous vehicles based on discrete global trajectory,

    Y . Liu, B. Zhou, and X. Wang, “Dynamic lane-changing trajectory planning for autonomous vehicles based on discrete global trajectory,” IEEE Transactions on Intelligent Transportation Systems, vol. 23, no. 7, pp. 8513–8527, 2022

    work page 2022

  17. [17]

    Safe and energy-efficient jerk- controlled speed profiling for on-road autonomous vehicles,

    F. Tarhini, R. Talj, and M. Doumiati, “Safe and energy-efficient jerk- controlled speed profiling for on-road autonomous vehicles,” IEEE Transactions on Intelligent Vehicles, pp. 1–16, 2024

    work page 2024

  18. [18]

    Specification-compliant driving corridors for motion planning of automated vehicles,

    E. I. Liu and M. Althoff, “Specification-compliant driving corridors for motion planning of automated vehicles,” IEEE Transactions on Intelligent Vehicles, vol. 8, no. 9, pp. 4180–4197, 2023

    work page 2023

  19. [19]

    Using reachable sets for trajec- tory planning of automated vehicles,

    S. Manzinger, C. Pek, and M. Althoff, “Using reachable sets for trajec- tory planning of automated vehicles,” IEEE Transactions on Intelligent Vehicles, vol. 6, no. 2, pp. 232–248, 2021

    work page 2021

  20. [20]

    Lateral and longitudinal integrated emer- gency collision avoidance system with handling stability guarantee,

    X. Chen, S. Cheng, and S. Li, “Lateral and longitudinal integrated emer- gency collision avoidance system with handling stability guarantee,” IEEE Transactions on Transportation Electrification, vol. 10, no. 3, pp. 4930–4940, 2024

    work page 2024

  21. [21]

    Stochastic model predictive control with a safety guarantee for automated driving,

    T. Brudigam, M. Olbrich, and D. Wollherr, “Stochastic model predictive control with a safety guarantee for automated driving,” IEEE Transac- tions on Intelligent Vehicles , vol. 8, no. 1, pp. 22–36, 2023. 14

    work page 2023

  22. [22]

    Interaction-aware decision-making for autonomous vehicles,

    Y . Chen, S. Li, and X. Tang, “Interaction-aware decision-making for autonomous vehicles,” IEEE Transactions on Transportation Electrifi- cation, vol. 9, no. 3, pp. 4704–4715, 2023

    work page 2023

  23. [23]

    Trajectory plan- ning in unknown 2d workspaces: A smooth, reactive, harmonics-based approach,

    P. Rousseas, C. Bechlioulis, and K. Kyriakopoulos, “Trajectory plan- ning in unknown 2d workspaces: A smooth, reactive, harmonics-based approach,” IEEE Robotics and Automation Letters , vol. 7, no. 2, pp. 1992–1999, 2022

    work page 1992

  24. [24]

    A motion planning and tracking framework for autonomous vehicles based on artificial potential field elaborated resistance network approach,

    Y . Huang, H. Ding, and Y . Zhang, “A motion planning and tracking framework for autonomous vehicles based on artificial potential field elaborated resistance network approach,” IEEE Transactions on Indus- trial Electronics, vol. 67, no. 2, pp. 1376–1386, 2020

    work page 2020

  25. [25]

    Toward safe motion planning for autonomous driving in highway,

    L. Cheng, Y . Qin, and K. Yang, “Toward safe motion planning for autonomous driving in highway,” IEEE Transactions on Vehicular Tech- nology, vol. 74, no. 2, pp. 2491–2502, Feb. 2025

    work page 2025

  26. [26]

    Interaction-aware motion planning for autonomous vehicles with multi-modal obstacle uncertainty predic- tions,

    J. Zhou, B. Olofsson, and E. Frisk, “Interaction-aware motion planning for autonomous vehicles with multi-modal obstacle uncertainty predic- tions,”IEEE Transactions on Intelligent Vehicles, vol. 9, no. 1, pp. 1305– 1319, Jan. 2024

    work page 2024

  27. [27]

    Rda: An accelerated collision free motion planner for autonomous navigation in cluttered environments,

    R. Han, S. Wang, and S. Wang, “Rda: An accelerated collision free motion planner for autonomous navigation in cluttered environments,” IEEE Robotics and Automation Letters , vol. 8, no. 3, pp. 1715–1722, 2023

    work page 2023

  28. [28]

    Computation of solution spaces for optimization-based trajectory planning,

    L. Schafer, S. Manzinger, and M. Althoff, “Computation of solution spaces for optimization-based trajectory planning,” IEEE Transactions on Intelligent Vehicles, vol. 8, no. 1, pp. 216–231, 2023

    work page 2023

  29. [29]

    End to End Learning for Self-Driving Cars

    M. Bojarski, D. Del Testa, D. Dworakowski, B. Firner, B. Flepp, P. Goyal, L. D. Jackel, M. Monfort, U. Muller, J. Zhang, X. Zhang, J. Zhao, and K. Zieba, “End to End Learning for Self-Driving Cars,” p. arXiv:1604.07316, Apr. 2016

    work page internal anchor Pith review Pith/arXiv arXiv 2016

  30. [30]

    Learning to drive in a day,

    A. Kendall, J. Hawke, D. Janz, P. Mazur, D. Reda, J.-M. Allen, V .-D. Lam, A. Bewley, and A. Shah, “Learning to drive in a day,” in 2019 International Conference on Robotics and Automation (ICRA), 2019, pp. 8248–8254

    work page 2019

  31. [31]

    Planning- oriented autonomous driving,

    Y . Hu, J. Yang, L. Chen, K. Li, C. Sima, X. Zhu, S. Chai, S. Du, T. Lin, W. Wang, L. Lu, X. Jia, Q. Liu, J. Dai, Y . Qiao, and H. Li, “Planning- oriented autonomous driving,” in Proceedings of the IEEE/CVF Confer- ence on Computer Vision and Pattern Recognition , 2023

    work page 2023

  32. [32]

    Combining decision making and trajectory planning for lane changing using deep reinforcement learning,

    S. Li, C. Wei, and Y . Wang, “Combining decision making and trajectory planning for lane changing using deep reinforcement learning,” IEEE Transactions on Intelligent Transportation Systems , vol. 23, no. 9, pp. 16 110–16 136, 2022

    work page 2022

  33. [33]

    Trajectory planning for autonomous driving in unstructured scenarios based on deep learning and quadratic optimization,

    H. Li, P. Chen, and G. Yu, “Trajectory planning for autonomous driving in unstructured scenarios based on deep learning and quadratic optimization,” IEEE Transactions on Vehicular Technology , vol. 73, no. 4, pp. 4886–4903, 2024

    work page 2024

  34. [34]

    Differentiable integrated motion predic- tion and planning with learnable cost function for autonomous driving,

    Z. Huang, H. Liu, and J. Wu, “Differentiable integrated motion predic- tion and planning with learnable cost function for autonomous driving,” IEEE Transactions on Neural Networks and Learning Systems , vol. 35, no. 11, pp. 15 222–15 236, Nov. 2024

    work page 2024

  35. [35]

    Optimal trajectory generation for dynamic street scenarios in a fren ´et frame,

    M. Werling, J. Ziegler, and S. Kammel, “Optimal trajectory generation for dynamic street scenarios in a fren ´et frame,” in 2010 IEEE Intelligent Vehicles Symposium, 2010, pp. 987–993

    work page 2010

  36. [36]

    Optimization, Gurobi Optimizer Reference Manual , 2024

    G. Optimization, Gurobi Optimizer Reference Manual , 2024

    work page 2024

  37. [37]

    The highd dataset: A drone dataset of naturalistic vehicle trajectories on german highways for validation of highly automated driving systems,

    R. Krajewski, J. Bock, L. Kloeker, and L. Eckstein, “The highd dataset: A drone dataset of naturalistic vehicle trajectories on german highways for validation of highly automated driving systems,” in 2018 21st IEEE International Conference on Intelligent Transportation Systems (ITSC) , Maui, Hawaii, USA, 2018, pp. 2118–2125

    work page 2018