pith. sign in

arxiv: 2605.06966 · v1 · submitted 2026-05-07 · 💻 cs.RO · cs.SE

Traffic Scenario Orchestration from Language via Constraint Satisfaction

Frieda Rong , Chris Zhang , Kelvin Wong , Raquel Urtasun This is my paper

Pith reviewed 2026-05-11 00:49 UTC · model grok-4.3

classification 💻 cs.RO cs.SE
keywords scenario generationconstraint satisfactionnatural language to constraintsautonomous vehicle testingclosed-loop simulationtraffic scenario orchestrationfoundation model reasoning
0
0 comments X

The pith

Natural language descriptions of traffic tests can be turned directly into constraint sets that off-the-shelf solvers use to generate closed-loop actor behaviors for autonomous vehicle evaluation.

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

The paper shows how to translate everyday language descriptions of complex driving situations into mathematical constraints that define required interactions among vehicles. These constraints are then handed to standard solvers that compute sequences of actions for all actors so the scene unfolds exactly as specified while the autonomous vehicle under test responds normally. This combination addresses the gap between manual scenario coding, which is precise but hard to scale, and statistical generators, which scale but lack controllability for out-of-distribution cases. The method records substantially higher success rates than baselines on a benchmark of diverse, precisely specified scenarios, with particular gains when the ego vehicle must react to other actors.

Core claim

The authors cast scenario orchestration as a constraint-solving problem. Foundation models convert general natural language descriptions into a set of constraints that serve as the scenario representation. Off-the-shelf solvers then find closed-loop trajectories for all actors that satisfy those constraints, meeting the precise testing intentions stated in the original language.

What carries the argument

The translation of natural language scenario descriptions into a constraint-based scenario representation that is solved by off-the-shelf solvers to produce closed-loop actor behaviors.

If this is right

  • Test engineers can specify interactive scenarios in plain English rather than writing custom code for each case.
  • Closed-loop solutions ensure other actors respond to the autonomous vehicle's actions, which is required for realistic ego-reactive testing.
  • The approach achieves higher orchestration success rates than prior baselines across diverse scenario descriptions.
  • Precise controllability becomes available for out-of-distribution cases that statistical models struggle to produce.

Where Pith is reading between the lines

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

  • The same language-to-constraint pipeline could be applied to generate test scenarios in other embodied simulation domains such as robotics or logistics.
  • If the constraint sets remain solvable at larger scale, the method could support automatic generation of thousands of targeted tests from a single high-level description.
  • Additional physical realism checks would be needed to filter mathematically valid but implausible solutions returned by the solver.

Load-bearing premise

Foundation models can reliably translate arbitrary natural language descriptions into accurate, complete, and solvable constraint sets without hallucinations, omissions, or inconsistencies.

What would settle it

A clear natural language description that produces a constraint set the solver either cannot satisfy or satisfies only with actor behaviors that contradict the original intent.

Figures

Figures reproduced from arXiv: 2605.06966 by Chris Zhang, Frieda Rong, Kelvin Wong, Raquel Urtasun.

Figure 1
Figure 1. Figure 1: We present a language-in, simulation-out framework [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: Our overall pipeline. A user inputs a scenario description in natural language, which gets converted by an LLM into [PITH_FULL_IMAGE:figures/full_fig_p002_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: While the solver reasons in 1-D by modelling the [PITH_FULL_IMAGE:figures/full_fig_p003_3.png] view at source ↗
Figure 5
Figure 5. Figure 5: Qualitative Comparison. We show the baseline (top) and our approach (bottom) on two different scenario descriptions. The baseline fails to accelerate past the conflict point. The line behind each actor indicates a 5s history.1 saved as template and reused as well. D. Closed-loop Replanning In most cases, the purpose of orchestration is to test the behavior an AV policy that is not under the control of the … view at source ↗
Figure 6
Figure 6. Figure 6: Side-by-side comparison of open loop vs. closed loop orchestration for a lead actor turn into driveway scenario. In [PITH_FULL_IMAGE:figures/full_fig_p006_6.png] view at source ↗
Figure 8
Figure 8. Figure 8: (left) Ours. (right) Baseline. The cell color represents [PITH_FULL_IMAGE:figures/full_fig_p006_8.png] view at source ↗
Figure 7
Figure 7. Figure 7: Importance of Closed-loop. We find closed-loop replanning in order to react to ego behaviors is important, especially for scenarios that require actor/ego interaction. (resp., time-to-arrival). We report success rate over two error tolerance levels. Overall, this metric allows us to measure the degree of controllability/precision a method is able to offer. B. Benchmarking Natural Language to Scenario We ev… view at source ↗
Figure 9
Figure 9. Figure 9: LLM Generation Runtime. We measure the average runtime of the one-time LLM generation phase. In our experiments, we use GPT-5 (medium reasoning effort). C. Precision & Controllability To assess the precision and controllability of our method, we analyze its performance on a fixed scenario across a discretized grid of parameter values in comparison to the baseline ( [PITH_FULL_IMAGE:figures/full_fig_p007_9.png] view at source ↗
read the original abstract

Autonomous vehicles (AVs) require extensive testing in simulation, but test case generation for driving scenarios is laborious. The desired scenarios are often out-of-distribution and have precise requirements on interactions with the AV policy under test. Manually programming scenarios allows for precise controllability but is difficult to scale. On the other hand, statistical models can leverage compute and data, but struggle with precise controllability when out-of-distribution. We cast scenario orchestration as a constraint-solving problem and present a language-in, simulation-out scenario orchestrator for closed-loop testing AVs. Our approach leverages foundation model reasoning to translate general, natural language descriptions into a set of constraints as a scenario representation. This then allows us to leverage off the shelf solvers to solve for actor behaviors which meet precise testing intentions in closed-loop. Under a benchmark of carefully crafted and diverse scenario descriptions, our approach greatly outperforms our baselines in orchestration success rate. We further show that our closed-loop approach is especially important for scenarios which require ego-reactive specifications.

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 / 2 minor

Summary. The paper presents a language-to-simulation pipeline for AV scenario orchestration: foundation models translate natural-language descriptions into constraint sets, which off-the-shelf solvers then use to generate closed-loop actor trajectories. The central claim is that this hybrid approach achieves substantially higher orchestration success rates than baselines on a benchmark of diverse, carefully crafted scenarios, with particular advantages for ego-reactive specifications.

Significance. If the generated constraints reliably and completely encode the original natural-language intent (without hallucinations, omissions, or inconsistencies that allow solver success on a divergent specification), the method would offer a scalable route to precise, out-of-distribution scenario generation that combines the controllability of manual programming with the flexibility of foundation models. The closed-loop emphasis and use of existing solvers are practical strengths.

major comments (2)
  1. [Abstract] Abstract and evaluation description: the reported 'orchestration success rate' is defined solely as the solver finding a feasible closed-loop solution. No post-generation validation is described that checks whether the solved trajectories satisfy the original natural-language description rather than a possibly hallucinated or incomplete constraint set. This directly undermines the claim that the method meets 'precise testing intentions.'
  2. [Abstract] Abstract: the outperformance claim is stated without any reported details on benchmark construction, baseline implementations, exact success metrics, or statistical significance testing. These omissions make it impossible to assess whether the central empirical result is robust.
minor comments (2)
  1. Clarify the precise interface between the foundation-model output and the solver input (e.g., how constraints are represented and whether they are checked for consistency before solving).
  2. The abstract mentions 'ego-reactive specifications' as a special case; a short example in the main text would help readers understand what distinguishes these from non-reactive cases.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the constructive feedback on our manuscript. We address each major comment point by point below. Where the comments identify areas for improvement in clarity or completeness, we have revised the manuscript accordingly.

read point-by-point responses

  1. Referee: [Abstract] Abstract and evaluation description: the reported 'orchestration success rate' is defined solely as the solver finding a feasible closed-loop solution. No post-generation validation is described that checks whether the solved trajectories satisfy the original natural-language description rather than a possibly hallucinated or incomplete constraint set. This directly undermines the claim that the method meets 'precise testing intentions.'

    Authors: We appreciate this observation. The orchestration success rate in the manuscript is defined as the solver returning a feasible closed-loop trajectory under the generated constraints, as this directly evaluates the end-to-end pipeline's ability to produce usable scenarios for AV testing. However, we acknowledge that this metric alone does not explicitly verify fidelity between the solved trajectories and the original natural-language intent. In the revised manuscript, we have added a post-generation validation procedure: an independent LLM-based checker verifies that key interaction elements from the original description are satisfied by the trajectories, supplemented by a human review on a random subset of scenarios. We have updated the abstract and evaluation sections to describe this validation and its results, thereby strengthening support for the claim of meeting precise testing intentions. revision: yes

  2. Referee: [Abstract] Abstract: the outperformance claim is stated without any reported details on benchmark construction, baseline implementations, exact success metrics, or statistical significance testing. These omissions make it impossible to assess whether the central empirical result is robust.

    Authors: We agree that the abstract's brevity omits these details. The full manuscript provides them in dedicated sections: benchmark construction (including scenario diversity and out-of-distribution focus) is described in Section 4; baseline implementations (with implementation specifics) appear in Section 5; exact success metrics (solver feasibility within time bounds) are defined in Section 3.2; and statistical significance (via paired t-tests with reported p-values) is included in the results tables. To address the concern, we have partially revised the abstract to briefly reference the benchmark's careful construction and the use of statistical testing, while retaining the original length constraints. This makes the empirical claims more transparent without changing the reported outcomes. revision: partial

Circularity Check

0 steps flagged

No circularity: external foundation models and solvers form an open derivation chain

full rationale

The paper's core pipeline translates natural-language scenario descriptions via an external foundation model into a constraint set, then invokes an off-the-shelf solver to produce closed-loop actor trajectories. No equation, definition, or success metric is shown to be constructed from the target result itself; the reported orchestration success rate is an external evaluation of solver feasibility on the generated constraints rather than a quantity fitted or renamed from the input descriptions. No self-citation chains, uniqueness theorems, or ansatzes imported from prior author work appear in the provided text as load-bearing steps. The derivation therefore remains self-contained against external components and does not reduce to its own inputs by construction.

Axiom & Free-Parameter Ledger

0 free parameters · 2 axioms · 0 invented entities

The central claim depends on the unverified capability of foundation models to produce complete and accurate constraint sets from natural language and on the existence of solvable constraint formulations for the described scenarios.

axioms (2)
  • domain assumption Foundation models can translate general natural language descriptions into accurate and complete constraint sets suitable for solvers
    Invoked in the description of the language-in, simulation-out pipeline.
  • domain assumption Off-the-shelf constraint solvers can find actor behaviors satisfying the generated constraints in closed-loop with an AV policy
    Core to the orchestration success claim.

pith-pipeline@v0.9.0 · 5470 in / 1386 out tokens · 61298 ms · 2026-05-11T00:49:20.777418+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

100 extracted references · 100 canonical work pages

  1. [1]

    ASAM OpenSCENARIO v2.0.0,

    ASAM e.V ., “ASAM OpenSCENARIO v2.0.0,” 2022, retrieved from https://www.asam.net/standards/detail/openscenario/v200/

    work page 2022

  2. [2]

    Language conditioned traffic generation,

    S. Tan, B. Ivanovic, X. Weng, M. Pavone, and P. Kraehenbuehl, “Language conditioned traffic generation,” in7th Annual Conference on Robot Learning, 2023. [Online]. Available: https://openreview.net/ forum?id=PK2debCKaG

    work page 2023

  3. [3]

    Promptable closed-loop traffic simulation,

    S. Tan, B. Ivanovic, Y . Chen, B. Li, X. Weng, Y . Cao, P. Kr ¨ahenb¨uhl, and M. Pavone, “Promptable closed-loop traffic simulation,” in8th Annual Conference on Robot Learning, 2024

    work page 2024

  4. [4]

    Geoscenario: An open dsl for autonomous driving scenario representation,

    R. Queiroz, T. Berger, and K. Czarnecki, “Geoscenario: An open dsl for autonomous driving scenario representation,” in2019 IEEE Intelligent Vehicles Symposium (IV). IEEE, 2019, pp. 287–294

    work page 2019

  5. [5]

    Scenic: a language for scenario specification and scene generation,

    D. J. Fremont, T. Dreossi, S. Ghosh, X. Yue, A. L. Sangiovanni- Vincentelli, and S. A. Seshia, “Scenic: a language for scenario specification and scene generation,” inProceedings of the 40th ACM SIGPLAN conference on programming language design and imple- mentation, 2019, pp. 63–78

    work page 2019

  6. [6]

    Metascenario: A framework for driving scenario data description, storage and indexing,

    C. Chang, D. Cao, L. Chen, K. Su, K. Su, Y . Su, F.-Y . Wang, J. Wang, P. Wang, J. Wei,et al., “Metascenario: A framework for driving scenario data description, storage and indexing,”IEEE Transactions on Intelligent Vehicles, vol. 8, no. 2, pp. 1156–1175, 2022

    work page 2022

  7. [7]

    Synthesizing traffic scenarios from formal specifications for testing automated vehicles,

    M. Klischat and M. Althoff, “Synthesizing traffic scenarios from formal specifications for testing automated vehicles,” in2020 IEEE intelligent vehicles symposium (IV). IEEE, 2020, pp. 2065–2072

    work page 2020

  8. [8]

    Crisgen: Constraint-based generation of critical scenarios for autonomous vehicles,

    A. Nonnengart, M. Klusch, and C. M ¨uller, “Crisgen: Constraint-based generation of critical scenarios for autonomous vehicles,” inFormal Methods. FM 2019 International Workshops: Porto, Portugal, October 7–11, 2019, Revised Selected Papers, Part I 3. Springer, 2020, pp. 233–248

    work page 2019

  9. [9]

    Solving constraint systems from traffic scenarios for the validation of autonomous driving

    K. Scheibler, A. Eggers, T. Teige, M. Walz, T. Bienm ¨uller, and U. Brockmeyer, “Solving constraint systems from traffic scenarios for the validation of autonomous driving.” inSC-square@ SIAM AG, 2019

    work page 2019

  10. [10]

    Sovar: Build generalizable scenarios from accident reports for autonomous driving testing,

    A. Guo, Y . Zhou, H. Tian, C. Fang, Y . Sun, W. Sun, X. Gao, A. T. Luu, Y . Liu, and Z. Chen, “Sovar: Build generalizable scenarios from accident reports for autonomous driving testing,” inProceedings of the 39th IEEE/ACM International Conference on Automated Software Engineering, 2024, pp. 268–280

    work page 2024

  11. [11]

    Foundation models in autonomous driving: A survey on scenario generation and scenario analysis,

    Y . Gao, M. Piccinini, Y . Zhang, D. Wang, K. Moller, R. Brusnicki, B. Zarrouki, A. Gambi, J. F. Totz, K. Storms,et al., “Foundation models in autonomous driving: A survey on scenario generation and scenario analysis,”arXiv preprint arXiv:2506.11526, 2025

    work page arXiv 2025

  12. [12]

    Text2scenario: Text-driven scenario generation for autonomous driving test,

    X. Cai, X. Bai, Z. Cui, D. Xie, D. Fu, H. Yu, and Y . Ren, “Text2scenario: Text-driven scenario generation for autonomous driv- ing test,”arXiv preprint arXiv:2503.02911, 2025

    work page arXiv 2025

  13. [13]

    Laser: Script execution by autonomous agents for on-demand traffic simulation,

    H. Gao, J. Wang, W. Fang, J. Xu, Y . Huang, T. Chen, and X. Ma, “Laser: Script execution by autonomous agents for on-demand traffic simulation,” inProceedings of the 16th International Conference on Internetware, 2025, pp. 84–95

    work page 2025

  14. [14]

    Legend: A top-down approach to scenario generation of autonomous driving systems assisted by large language models,

    S. Tang, Z. Zhang, J. Zhou, L. Lei, Y . Zhou, and Y . Xue, “Legend: A top-down approach to scenario generation of autonomous driving systems assisted by large language models,” inProceedings of the 39th IEEE/ACM International Conference on Automated Software Engineering, 2024, pp. 1497–1508

    work page 2024

  15. [15]

    Generating probabilistic scenario programs from natural language,

    K. Elmaaroufi, D. Shanker, A. Cismaru, M. Vazquez-Chanlatte, A. L. Sangiovanni-Vincentelli, M. Zaharia, and S. A. Seshia, “Generating probabilistic scenario programs from natural language,”CoRR, 2024

    work page 2024

  16. [16]

    Text-to-drive: Diverse driving behavior synthesis via large language models,

    P. Nguyen, T.-H. Wang, Z.-W. Hong, S. Karaman, and D. Rus, “Text-to-drive: Diverse driving behavior synthesis via large language models,” 2024

    work page 2024

  17. [17]

    Spacetime constraints,

    A. Witkin and M. Kass, “Spacetime constraints,”ACM Siggraph Computer Graphics, vol. 22, no. 4, pp. 159–168, 1988

    work page 1988

  18. [18]

    Automated synthesis of numer- ical programs for simulation of rigid mechanical systems in physics- based animation,

    T. Ellman, R. Deak, and J. Fotinatos, “Automated synthesis of numer- ical programs for simulation of rigid mechanical systems in physics- based animation,”Automated Software Engineering, vol. 10, pp. 367– 398, 2003

    work page 2003

  19. [19]

    OpenAI, “Gpt-5,” https://platform.openai.com, 2025, openAI API

    work page 2025

  20. [20]

    Z3: An efficient smt solver,

    L. De Moura and N. Bjørner, “Z3: An efficient smt solver,” in International conference on Tools and Algorithms for the Construction and Analysis of Systems. Springer, 2008, pp. 337–340. VI. SCENARIOTAXONOMY Ego State Ego Intent Hero State Hero Intent Interaction Trigger Driveway Right turn Driveway Right turn Hero brakes Distance Driveway Right turn Westb...

    work page 2008

  21. [21]

    It wants to turn right onto the road travelling westbound

    Ego is on the driveway exit and it is initial ego distance behind stop line m behind the stop line. It wants to turn right onto the road travelling westbound. There is a hero vehicle on the driveway exit and it is ini- tial hero distance behind stop line m behind the stop line. It wants to turn right onto the road travelling westbound. When the ego is ego...

    work page

  22. [22]

    It wants to turn right onto the road travelling westbound

    Ego is on the driveway exit and it is initial ego distance behind stop line m behind the stop line. It wants to turn right onto the road travelling westbound. There is a hero vehicle on the road. It wants to lane follow along the road travelling westbound. When the ego is ego distance behind stop line m behind the stop line and the hero vehicle is hero di...

    work page

  23. [23]

    It wants to turn right onto the road travelling westbound

    Ego is on the driveway exit and it is initial ego distance behind stop line m behind the stop line. It wants to turn right onto the road travelling westbound. There is a hero vehicle on the road. It wants to lane follow along the road travelling westbound. When the ego is ego distance behind stop line m behind the stop line and the hero vehicle is distanc...

    work page

  24. [24]

    It wants to turn right onto the road travelling westbound

    Ego is on the driveway exit and it is initial ego distance behind stop line m behind the stop line. It wants to turn right onto the road travelling westbound. There is a hero vehicle on the road. It wants to lane follow along the road travelling westbound. When the ego is ego distance behind stop line m behind the stop line and the hero vehicle is distanc...

    work page

  25. [25]

    It wants to turn left onto the road travelling eastbound

    Ego is on the driveway exit and it is initial ego distance behind stop line m behind the stop line. It wants to turn left onto the road travelling eastbound. There is a hero vehicle on the driveway exit and it is ini- tial hero distance behind stop line m behind the stop line. It wants to turn left onto the road travelling eastbound. When the ego is ego d...

    work page

  26. [26]

    It wants to turn left onto the road travelling eastbound

    Ego is on the driveway exit and it is initial ego distance behind stop line m behind the stop line. It wants to turn left onto the road travelling eastbound. There is a hero vehicle on the road. It wants to lane follow along the road travelling westbound. When the ego is ego distance behind stop line m behind the stop line and the hero vehicle is hero dis...

    work page

  27. [27]

    It wants to turn left onto the road travelling eastbound

    Ego is on the driveway exit and it is initial ego distance behind stop line m behind the stop line. It wants to turn left onto the road travelling eastbound. There is a hero vehicle on the road. It wants to lane follow along the road travelling westbound. When the ego is ego distance behind stop line m behind the stop line and the hero vehicle is distance...

    work page

  28. [28]

    It wants to turn left onto the road travelling eastbound

    Ego is on the driveway exit and it is initial ego distance behind stop line m behind the stop line. It wants to turn left onto the road travelling eastbound. There is a hero vehicle on the road. It wants to lane follow along the road travelling westbound. When the ego is ego distance behind stop line m behind the stop line and the hero vehicle is distance...

    work page

  29. [29]

    It wants to turn left onto the road travelling eastbound

    Ego is on the driveway exit and it is initial ego distance behind stop line m behind the stop line. It wants to turn left onto the road travelling eastbound. There is a hero vehicle on the road. It wants to lane follow along the road travelling eastbound. When the ego is ego distance behind stop line m behind the stop line and the hero vehicle is hero dis...

    work page

  30. [30]

    It wants to turn left onto the road travelling eastbound

    Ego is on the driveway exit and it is initial ego distance behind stop line m behind the stop line. It wants to turn left onto the road travelling eastbound. There is a hero vehicle on the road. It wants to lane follow along the road travelling eastbound. When the ego is ego distance behind stop line m behind the stop line and the hero vehicle is distance...

    work page

  31. [31]

    It wants to turn left onto the road travelling eastbound

    Ego is on the driveway exit and it is initial ego distance behind stop line m behind the stop line. It wants to turn left onto the road travelling eastbound. There is a hero vehicle on the road. It wants to lane follow along the road travelling eastbound. When the ego is ego distance behind stop line m behind the stop line and the hero vehicle is distance...

    work page

  32. [32]

    It wants to turn left onto the road travelling eastbound

    Ego is on the driveway exit and it is initial ego distance behind stop line m behind the stop line. It wants to turn left onto the road travelling eastbound. There is a hero vehicle on the road. It wants to turn left into the driveway. When the ego is ego distance behind stop line m behind the stop line and the hero vehicle is hero distance behind conflic...

    work page

  33. [33]

    It wants to turn left onto the road travelling eastbound

    Ego is on the driveway exit and it is initial ego distance behind stop line m behind the stop line. It wants to turn left onto the road travelling eastbound. There is a hero vehicle on the road. It wants to turn left into the driveway. When the ego is ego distance behind stop line m behind the stop line and the hero vehicle is distance behind conflict poi...

    work page

  34. [34]

    It wants to turn left onto the road travelling eastbound

    Ego is on the driveway exit and it is initial ego distance behind stop line m behind the stop line. It wants to turn left onto the road travelling eastbound. There is a hero vehicle on the road. It wants to turn left into the driveway. When the ego is ego distance behind stop line m behind the stop line and the hero vehicle is distance behind conflict poi...

    work page

  35. [35]

    It wants to lane follow along the road travelling westbound

    Ego is on the road and it is initial ego distance behind intersection m behind the intersection. It wants to lane follow along the road travelling westbound. There is a hero vehicle on the road. It wants to lane follow along the road travelling westbound. When the ego is ego distance behind hero m behind the hero vehicle, the hero vehicle begins to decele...

    work page

  36. [36]

    It wants to lane follow along the road travelling westbound

    Ego is on the road and it is initial ego distance behind intersection m behind the intersection. It wants to lane follow along the road travelling westbound. There is a hero vehicle on the road. It wants to turn left into the driveway. When the ego is ego distance behind conflict point m behind the the conflict point and the hero vehicle is hero distance ...

    work page

  37. [37]

    It wants to lane follow along the road travelling westbound

    Ego is on the road and it is initial ego distance behind intersection m behind the intersection. It wants to lane follow along the road travelling westbound. There is a hero vehicle on the road. It wants to turn left into the driveway. When the ego is ego distance behind conflict point m behind the conflict point and the hero vehicle is hero distance behi...

    work page

  38. [38]

    It wants to lane follow along the road travelling westbound

    Ego is on the road and it is initial ego distance behind intersection m behind the intersection. It wants to lane follow along the road travelling westbound. There is a hero vehicle on the road. It wants to turn left into the driveway. When the ego is ego distance behind conflict point m behind the conflict point and the hero vehicle is hero distance behi...

    work page

  39. [39]

    It wants to lane follow along the road travelling westbound

    Ego is on the road and it is initial ego distance behind intersection m behind the intersection. It wants to lane follow along the road travelling westbound. There is a hero vehicle on the driveway. It wants to turn right onto the road travelling westbound. When the ego is ego distance behind conflict point m behind the conflict point and the hero vehicle...

    work page

  40. [40]

    It wants to lane follow along the road travelling westbound

    Ego is on the road and it is initial ego distance behind intersection m behind the intersection. It wants to lane follow along the road travelling westbound. There is a hero vehicle on the driveway. It wants to turn right onto the road travelling westbound. When the ego is ego distance behind conflict point m behind the conflict point and the hero vehicle...

    work page

  41. [41]

    It wants to lane follow along the road travelling westbound

    Ego is on the road and it is initial ego distance behind intersection m behind the intersection. It wants to lane follow along the road travelling westbound. There is a hero vehicle on the driveway. It wants to turn right onto the road travelling westbound. When the ego is ego distance behind conflict point m behind the conflict point and the hero vehicle...

    work page

  42. [42]

    It wants to lane follow along the road travelling westbound

    Ego is on the road and it is initial ego distance behind intersection m behind the intersection. It wants to lane follow along the road travelling westbound. There is a hero vehicle on the driveway. It wants to turn left onto the road travelling eastbound. When the ego is ego distance behind conflict point m behind the conflict point and the hero vehicle ...

    work page

  43. [43]

    It wants to lane follow along the road travelling westbound

    Ego is on the road and it is initial ego distance behind intersection m behind the intersection. It wants to lane follow along the road travelling westbound. There is a hero vehicle on the driveway. It wants to turn left onto the road travelling eastbound. When the ego is ego distance behind conflict point m behind the conflict point and the hero vehicle ...

    work page

  44. [44]

    It wants to lane follow along the road travelling westbound

    Ego is on the road and it is initial ego distance behind intersection m behind the intersection. It wants to lane follow along the road travelling westbound. There is a hero vehicle on the driveway. It wants to turn left onto the road travelling eastbound. When the ego is ego distance behind conflict point m behind the conflict point and the hero vehicle ...

    work page

  45. [45]

    It wants to turn right into the driveway

    Ego is on the road and it is initial ego distance behind intersection m behind the intersection. It wants to turn right into the driveway. There is a hero vehicle on the road. It wants to turn left into the drive- way. When the ego is ego distance behind conflict point m behind the conflict point and the hero vehicle is hero distance behind conflict point...

    work page

  46. [46]

    It wants to lane follow along the road travelling westbound

    Ego is on the road and it is initial ego distance behind intersection m behind the intersection. It wants to lane follow along the road travelling westbound. There is a hero vehicle on the road. It wants to turn left into the driveway. When the ego is ego distance behind conflict point m behind the conflict point and the hero vehicle is hero distance behi...

    work page

  47. [47]

    It wants to lane follow along the road travelling westbound

    Ego is on the road and it is initial ego distance behind intersection m behind the intersection. It wants to lane follow along the road travelling westbound. There is a hero vehicle on the road. It wants to turn left into the driveway. When the ego is ego distance behind conflict point m behind the conflict point and the hero vehicle is distance behind co...

    work page

  48. [48]

    It wants to lane follow along the road travelling eastbound

    Ego is on the road and it is initial ego distance behind intersection m behind the intersection. It wants to lane follow along the road travelling eastbound. There is a hero vehicle on the road. It wants to lane follow along the road travelling eastbound. When the ego is ego distance behind hero m behind the hero vehicle, the hero vehicle begins to decele...

    work page

  49. [49]

    It wants to lane follow along the road travelling eastbound

    Ego is on the road and it is initial ego distance behind intersection m behind the intersection. It wants to lane follow along the road travelling eastbound. There is a hero vehicle on the driveway. It wants to turn left onto the road travelling eastbound. When the ego is ego distance behind conflict point m behind the conflict point and the hero vehicle ...

    work page

  50. [50]

    It wants to lane follow along the road travelling eastbound

    Ego is on the road and it is initial ego distance behind intersection m behind the intersection. It wants to lane follow along the road travelling eastbound. There is a hero vehicle on the driveway. It wants to turn left onto the road travelling eastbound. When the ego is ego distance behind conflict point m behind the conflict point and the hero vehicle ...

    work page

  51. [51]

    It wants to lane follow along the road travelling eastbound

    Ego is on the road and it is initial ego distance behind intersection m behind the intersection. It wants to lane follow along the road travelling eastbound. There is a hero vehicle on the driveway. It wants to turn left onto the road travelling eastbound. When the ego is ego distance behind conflict point m behind the conflict point and the hero vehicle ...

    work page

  52. [52]

    It wants to turn left into the driveway

    Ego is on the road and it is initial ego distance behind intersection m behind the intersection. It wants to turn left into the driveway. There is a hero vehicle on the road. It wants to lane follow along the road travelling westbound. When the ego is ego distance behind conflict point m behind the conflict point and the hero vehicle is hero distance behi...

    work page

  53. [53]

    It wants to turn left into the driveway

    Ego is on the road and it is initial ego distance behind intersection m behind the intersection. It wants to turn left into the driveway. There is a hero vehicle on the road. It wants to lane follow along the road travelling westbound. When the ego is ego distance behind conflict point m behind the conflict point and the hero vehicle is distance behind co...

    work page

  54. [54]

    It wants to turn left into the driveway

    Ego is on the road and it is initial ego distance behind intersection m behind the intersection. It wants to turn left into the driveway. There is a hero vehicle on the road. It wants to lane follow along the road travelling westbound. When the ego is ego distance behind conflict point m behind the conflict point and the hero vehicle is distance behind co...

    work page

  55. [55]

    It wants to turn left into the driveway

    Ego is on the road and it is initial ego distance behind intersection m behind the intersection. It wants to turn left into the driveway. There is a hero vehicle on the road. It wants to turn right into the drive- way. When the ego is ego distance behind conflict point m behind the conflict point and the hero vehicle is hero distance behind conflict point...

    work page

  56. [56]

    It wants to turn left into the driveway

    Ego is on the road and it is initial ego distance behind intersection m behind the intersection. It wants to turn left into the driveway. There is a hero vehicle on the road. It wants to turn right into the driveway. When the ego is ego distance behind conflict point m behind the conflict point and the hero vehicle is distance behind conflict point m behi...

    work page

  57. [57]

    It wants to turn left into the driveway

    Ego is on the road and it is initial ego distance behind intersection m behind the intersection. It wants to turn left into the driveway. There is a hero vehicle on the road. It wants to turn right into the driveway. When the ego is ego distance behind conflict point m behind the conflict point and the hero vehicle is dis- tance behind conflict point m be...

    work page

  58. [58]

    It wants to turn left into the driveway

    Ego is on the road and it is initial ego distance behind intersection m behind the intersection. It wants to turn left into the driveway. There is a hero vehicle on the driveway. It wants to turn left onto the road travelling eastbound. When the ego is ego distance behind conflict point m behind the conflict point and the hero vehicle is hero distance beh...

    work page

  59. [59]

    It wants to turn left into the driveway

    Ego is on the road and it is initial ego distance behind intersection m behind the intersection. It wants to turn left into the driveway. There is a hero vehicle on the driveway. It wants to turn left onto the road travelling eastbound. When the ego is ego distance behind conflict point m behind the conflict point and the hero vehicle is hero distance beh...

    work page

  60. [60]

    It wants to turn left into the driveway

    Ego is on the road and it is initial ego distance behind intersection m behind the intersection. It wants to turn left into the driveway. There is a hero vehicle on the driveway. It wants to turn left onto the road travelling eastbound. When the ego is ego distance behind conflict point m behind the conflict point and the hero vehicle is hero distance beh...

    work page

  61. [61]

    It wants to turn right onto the road travelling westbound

    Ego is on the driveway exit and it is initial ego tta behind stop line s from the stop line. It wants to turn right onto the road travelling westbound. There is a hero vehicle on the driveway exit and it is initial hero tta behind stop line s from the stop line. It wants to turn right onto the road travelling westbound. When the ego is ego distance behind...

    work page

  62. [62]

    It wants to turn right onto the road travelling westbound

    Ego is on the driveway exit and it is initial ego tta behind stop line s from the stop line. It wants to turn right onto the road travelling westbound. There is a hero vehicle on the road. It wants to lane follow along the road travelling westbound. When the ego is ego tta behind stop line s from the stop line and the hero vehicle is hero tta behind confl...

    work page

  63. [63]

    It wants to turn right onto the road travelling westbound

    Ego is on the driveway exit and it is initial ego tta behind stop line s from the stop line. It wants to turn right onto the road travelling westbound. There is a hero vehicle on the road. It wants to lane follow along the road travelling westbound. When the ego is ego tta behind stop line s from the stop line and the hero vehicle is tta behind conflict p...

    work page

  64. [64]

    It wants to turn right onto the road travelling westbound

    Ego is on the driveway exit and it is initial ego tta behind stop line s from the stop line. It wants to turn right onto the road travelling westbound. There is a hero vehicle on the road. It wants to lane follow along the road travelling westbound. When the ego is ego tta behind stop line s from the stop line and the hero vehicle is tta behind conflict p...

    work page

  65. [65]

    It wants to turn left onto the road travelling eastbound

    Ego is on the driveway exit and it is initial ego tta behind stop line s from the stop line. It wants to turn left onto the road travelling eastbound. There is a hero vehicle on the driveway exit and it is initial hero tta behind stop line s from the stop line. It wants to turn left onto the road travelling eastbound. When the ego is ego distance behind h...

    work page

  66. [66]

    It wants to turn left onto the road travelling eastbound

    Ego is on the driveway exit and it is initial ego tta behind stop line s from the stop line. It wants to turn left onto the road travelling eastbound. There is a hero vehicle on the road. It wants to lane follow along the road travelling westbound. When the ego is ego tta behind stop line s from the stop line and the hero vehicle is hero tta behind confli...

    work page

  67. [67]

    It wants to turn left onto the road travelling eastbound

    Ego is on the driveway exit and it is initial ego tta behind stop line s from the stop line. It wants to turn left onto the road travelling eastbound. There is a hero vehicle on the road. It wants to lane follow along the road travelling westbound. When the ego is ego tta behind stop line s from the stop line and the hero vehicle is tta behind conflict po...

    work page

  68. [68]

    It wants to turn left onto the road travelling eastbound

    Ego is on the driveway exit and it is initial ego tta behind stop line s from the stop line. It wants to turn left onto the road travelling eastbound. There is a hero vehicle on the road. It wants to lane follow along the road travelling westbound. When the ego is ego tta behind stop line s from the stop line and the hero vehicle is tta behind conflict po...

    work page

  69. [69]

    It wants to turn left onto the road travelling eastbound

    Ego is on the driveway exit and it is initial ego tta behind stop line s from the stop line. It wants to turn left onto the road travelling eastbound. There is a hero vehicle on the road. It wants to lane follow along the road travelling eastbound. When the ego is ego tta behind stop line s from the stop line and the hero vehicle is hero tta behind confli...

    work page

  70. [70]

    It wants to turn left onto the road travelling eastbound

    Ego is on the driveway exit and it is initial ego tta behind stop line s from the stop line. It wants to turn left onto the road travelling eastbound. There is a hero vehicle on the road. It wants to lane follow along the road travelling eastbound. When the ego is ego tta behind stop line s from the stop line and the hero vehicle is tta behind conflict po...

    work page

  71. [71]

    It wants to turn left onto the road travelling eastbound

    Ego is on the driveway exit and it is initial ego tta behind stop line s from the stop line. It wants to turn left onto the road travelling eastbound. There is a hero vehicle on the road. It wants to lane follow along the road travelling eastbound. When the ego is ego tta behind stop line s from the stop line and the hero vehicle is tta behind conflict po...

    work page

  72. [72]

    It wants to turn left onto the road travelling eastbound

    Ego is on the driveway exit and it is initial ego tta behind stop line s from the stop line. It wants to turn left onto the road travelling eastbound. There is a hero vehicle on the road. It wants to turn left into the driveway. When the ego is ego tta behind stop line s from the stop line and the hero vehicle is hero tta behind conflict point s from the ...

    work page

  73. [73]

    It wants to turn left onto the road travelling eastbound

    Ego is on the driveway exit and it is initial ego tta behind stop line s from the stop line. It wants to turn left onto the road travelling eastbound. There is a hero vehicle on the road. It wants to turn left into the driveway. When the ego is ego tta behind stop line s from the stop line and the hero vehicle is tta behind conflict point s from the confl...

    work page

  74. [74]

    It wants to turn left onto the road travelling eastbound

    Ego is on the driveway exit and it is initial ego tta behind stop line s from the stop line. It wants to turn left onto the road travelling eastbound. There is a hero vehicle on the road. It wants to turn left into the driveway. When the ego is ego tta behind stop line s from the stop line and the hero vehicle is tta behind conflict point s from the confl...

    work page

  75. [75]

    It wants to lane follow along the road travelling westbound

    Ego is on the road and it is initial ego tta behind intersection s from the intersection. It wants to lane follow along the road travelling westbound. There is a hero vehicle on the road. It wants to lane follow along the road travelling westbound. When the ego is ego distance behind hero m behind the hero vehicle, the hero vehicle begins to decelerate to...

    work page

  76. [76]

    It wants to lane follow along the road travelling westbound

    Ego is on the road and it is initial ego tta behind intersection s from the intersection. It wants to lane follow along the road travelling westbound. There is a hero vehicle on the road. It wants to turn left into the driveway. When the ego is ego tta behind conflict point s from the conflict point and the hero vehicle is hero tta behind conflict point s...

    work page

  77. [77]

    It wants to lane follow along the road travelling westbound

    Ego is on the road and it is initial ego tta behind intersection s from the intersection. It wants to lane follow along the road travelling westbound. There is a hero vehicle on the road. It wants to turn left into the driveway. When the ego is ego tta behind conflict point s from the conflict point and the hero vehicle is hero tta behind conflict point s...

    work page

  78. [78]

    It wants to lane follow along the road travelling westbound

    Ego is on the road and it is initial ego tta behind intersection s from the intersection. It wants to lane follow along the road travelling westbound. There is a hero vehicle on the road. It wants to turn left into the driveway. When the ego is ego tta behind conflict point s from the conflict point and the hero vehicle is hero tta behind conflict point s...

    work page

  79. [79]

    It wants to lane follow along the road travelling westbound

    Ego is on the road and it is initial ego tta behind intersection s from the intersection. It wants to lane follow along the road travelling westbound. There is a hero vehicle on the driveway. It wants to turn right onto the road travelling westbound. When the ego is ego tta behind conflict point s from the conflict point and the hero vehicle is hero tta b...

    work page

  80. [80]

    It wants to lane follow along the road travelling westbound

    Ego is on the road and it is initial ego tta behind intersection s from the intersection. It wants to lane follow along the road travelling westbound. There is a hero vehicle on the driveway. It wants to turn right onto the road travelling westbound. When the ego is ego tta behind conflict point s from the conflict point and the hero vehicle is hero tta b...

    work page

Showing first 80 references.