Pith. sign in

REVIEW 1 major objections 5 minor 37 references

When Shared Rollouts Fail in Defensive Driving Evaluation: A NAVSIM Score Basis Audit

T0 review · 1 major / 5 minor · reviewed 2026-08-06 · deepseek-v4-flash

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

Pith's one-line read A documented NAVSIM scorer configuration ranks a route-blind policy above human replay and PDM-Closed on the full navtest split; the paper traces this to a numerical instability that forgiveness turns into compliance credit.

desk verdict A careful, honestly scoped audit showing that a shared velocity-refit instability plus reference-conditioned forgiveness can invert NAVSIM rankings; the causal 'direct trigger' claim is slightly ahead of the evidence on the full navtest split, but the paper deserves peer review. read the letter →

arxiv 2608.04896 v1 pith:XRWQPOGU submitted 2026-08-05 cs.AI cs.CV

Ziang Wei , Minjun Yu , Zheyuan Lai , Mingjie Pang , Wei Li This is my paper · ORCID
classification cs.AIcs.CV
keywords defensivedrivingevaluationbenchmarkauditingreference-conditionedforgivenessactor-blindprobesNAVSIMnumericalstabilityshared-failurecollapserollout
verification ladder T0 review T1 audit T2 compute T3 formal

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 argues that a defensive-driving score is only meaningful if it keeps actor-aware policies above actor-blind ones, and then shows a documented NAVSIM scorer configuration violates that condition. Under the affected numerical stack, a route-blind policy that ignores all surrounding actors scores 79.6 EPDMS, outranking human replay (74.0), DiffusionDrive (73.7), and PDM-Closed (67.1) over the complete 12,146-token navtest split. The cause is a dependency-sensitive instability in the shared velocity refit: both the agent proposal and the logged human reference traverse the same transformation, and when the reference rollout degenerates, the scorer's reference-conditioned forgiveness—which grants credit whenever the logged human reference fails a compliance channel—overwrites the agent's compliance failures into passes. Replacing the solver or using a same-source dependency stack removes the divergence and restores blind-last ordering on the 450-token control pool, so the paper concludes that the numerical instability is the direct trigger and forgiveness is the amplifier. The upshot is an audit protocol—dependency and backend disclosure, blind probes, overwrite reporting, and rollout stability tests—for any re-simulation score used to support defensive-driving claims.

What carries the argument

The load-bearing mechanism is shared-failure forgiveness collapse, which needs three ingredients: a transformation $T(\cdot, s)$ applied to both the agent proposal $x_a$ and the logged human reference $x_r$; a reference failure mask $F_j(x_r,s) = \mathbb{1}[m_j(T(x_r, s), s) = 0]$ for each compliance channel $j$; and a reference-conditioned scoring rule that sets $\tilde{m}_j(x_a, x_r, s) = 1$ whenever $F_j(x_r, s) = 1$ and otherwise uses $m_j(T(x_a, s), s)$. When the shared refit is numerically unstable, both sides fail the structural channels, so the masks convert almost every compliance failure into a pass and leave only residual terms to determine the ranking. The diagnostics that carry the argument are the stock filter-on versus filter-off comparison, a two-sided finite-difference diagnostic that bypasses the unstable refit on both sides, solver replacement with a direct solve or Hermitian pseudoinverse, and a same-source dependency stack control.

What would settle it

Run the solver-replacement control and the same-source dependency stack control on the full 12,146-token navtest split with forgiveness enabled, and compare per-token rollouts and final EPDMS scores: if Ignore-All falls below human replay and PDM-Closed and the divergence rate drops to near zero, the instability is confirmed as the direct trigger, while if the blind win persists despite the stabilized shared refit, the causal conclusion is incomplete.

Watch

Extended reading notes

Core claim

The paper's central claim is that, in the audited NAVSIM v2.2 original-scene single-stage scorer under the affected documented-stack condition, reference-conditioned forgiveness plus an unstable shared rollout transformation produces what the authors call shared-failure forgiveness collapse. Concretely, the default pseudoinverse in the velocity refit can produce unphysical profiles (median simulated endpoint of 5.13 km on a fixed 32-token diagnostic set), the logged human reference fails the same compliance channels, and the scorer then overwrites both sides' failures to passes in Drivable Area Compliance, Driving Direction Compliance, and lane keeping on roughly 99.9% of audited tokens. The result is that the route-blind Ignore-All probe (79.6 EPDMS) and a route-aware actor-blind probe (79.2) outrank human replay (74.0), DiffusionDrive (73.7), and PDM-Closed (67.1) on the full 12,146-token navtest split. Because the ordering reverses under a two-sided finite-difference diagnostic and on 450-token control pools when only the solver is replaced, the paper concludes that the numerical instability is the direct trigger, not a flaw in forgiveness alone.

Load-bearing premise

The load-bearing assumption is that the causal diagnosis from the 450-token navhard stage-1 control pool and the 32-token diagnostic set transfers to the full 12,146-token navtest split; the paper does not run the solver-replacement or same-source stack control on the full split, so an additional cause of the navtest anomaly cannot be excluded.

Editorial extensions

If this is right

  • Published NAVSIM-style rankings produced under the affected stack can invert: a policy that receives no surrounding-actor input can top actor-aware comparators, so such scores do not by themselves support actor-sensitive defensive-driving claims.
  • The fail-to-pass overwrite is concentrated in structural channels—DAC, DDC, and lane keeping—while TTC and no-at-fault collision are barely touched, so the collapse removes the very constraints that would penalize non-compliant trajectories.
  • Stabilizing the shared rollout, either by replacing the solver or by using a same-source dependency stack, removes divergence and restores blind-last ordering while keeping forgiveness enabled, showing the instability, not the forgiveness rule, is the direct trigger.
  • Reference-conditioned forgiveness remains useful against genuinely bad references, but it needs rollout stability checks, since it has no built-in way to tell a shared numerical failure from a legitimate reference exception.
  • Any re-simulation score used for defensive-driving claims should report the full score basis: dependency versions, numerical backend, reference failure masks, overwrite rates, and fitted-state and rollout stability, plus route-blind and route-aware actor-blind probes.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • If other re-simulation benchmarks share the same structure—a shared rollout transform plus reference-conditioned forgiveness—the collapse could reproduce there; adding a fixed numerical-stability diagnostic to their test suites would catch it cheaply.
  • Running the solver-replacement control on the full 12,146-token navtest split would settle whether the numerical instability is the whole story or only part of it; the paper leaves that run undone.
  • A practical implication the authors leave implicit is that benchmark leaders could require a score-basis attestation—exact dependency versions, backend, and thresholds—as a condition of submission, making ranking anomalies auditable.
  • The route-aware actor-blind probe could be reused as a standard sanity-check baseline in any planner benchmark, since it separates route-following competence from actor-awareness without retuning.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

1 major / 5 minor

Summary. This paper audits NAVSIM v2.2 original-scene single-stage scoring. It demonstrates that under a specific documented dependency stack (commit 0a380a9, OpenBLAS, NumPy 1.23.4, SciPy 1.13.1), two actor-blind probes—the route-blind Ignore-All and a route-aware actor-blind probe—outrank human replay and PDM-Closed on the complete 12,146-token navtest split. The paper attributes the reversal to shared-failure forgiveness collapse: the logged human reference and the agent share the same unstable velocity-refit rollout; when the reference fails compliance channels, reference-conditioned forgiveness rewrites agent failures to passes. It supports this claim with ingredient ablation, solver replacement on a 450-token control pool, a fresh-install 32-token reproduction, an exact-input 40x40 diagnostic, and a two-sided finite-difference diagnostic that restores blind-last ordering. The paper also contributes a reusable audit protocol requiring score-basis and stack disclosure, blind probes, overwrite reporting, and rollout stability tests.

Significance. If the findings hold, they identify a concrete and reproducible failure mode in a widely used re-simulation benchmark: a dependency-sensitive numerical instability in a shared planner/reference transform can invert the intended ranking of actor-aware versus actor-blind policies. The paper's strengths are its transparency (frozen artifacts, SHA-256 hashed exact inputs, token lists, per-token outputs), its multi-layered controls (filter-off, solver replacement, same-source stack, two-sided diagnostic), and its honest scoping of the 32-token diagnostic as mechanism evidence rather than prevalence. The audit protocol itself is a useful contribution. The main weakness is the gap between the scoped solver-replacement evidence and the full-split causal claim, which the abstract states more strongly than the evidence supports.

major comments (1)
  1. [Abstract and §5 (with Tables 5–7 and §6 Limitations)] The abstract states that 'the numerical instability is the direct trigger' of the full 12,146-token navtest reversal (Ignore-All 79.6 vs. human replay 74.0, Table 2). The minimal solver control that isolates this trigger is run on the 450-token navhard stage-1 pool (Table 5) and the 32-token diagnostic set (Table 6), not on the navtest split where the headline reversal is measured. On the full split, the two-sided finite-difference diagnostic (Table 7) and the fallback guard are broader interventions that change the score basis beyond the minimal velocity-refit solver replacement. The Limitations section (Sec. 6) correctly labels these controls as 'scoped,' but the abstract and §5 do not carry the same caveat. It therefore remains open whether the navtest reversal is fully explained by the refit instability alone, or whether a second dependency-sensitive failure on the navtest distribution, or a representativeness gap between the navhard control pool and navtest, contributes to the 79.6 vs. 74.0 margin. Since the full token list and stack are provided, running the solver replacement (direct solve or Hermitian pseudoinverse) on the full navtest split appears feasible and would directly secure the causal claim; otherwise the 'direct trigger' wording should be relaxed to 'the direct trigger on the audited control pools.'
minor comments (5)
  1. [Title and throughout] The title contains a spacing typo: 'NA VSIM' should be 'NAVSIM'; similar spacing issues appear in table cells such as 'Ignore-All79.6' and 'Turn+ moving' in Tables 2–8.
  2. [§3.3] The scoring notation is inconsistent: the text writes m_j(T(x), s) but the definition of T in the preceding line suggests T takes (x, s). Please clarify the argument structure, e.g., write T(x, s) or define the composition explicitly.
  3. [§4.4] The claim that 'no convex reweighting of the component scores gives PDM-Closed more than a 0.17-point margin over Ignore-All' is not accompanied by a definition of the component score set or the reweighting constraints; please add a footnote or supplementary detail describing the optimization and the components included.
  4. [§4.2] The term 'BatchLQR' is introduced without definition; please state that it is the velocity/curvature refit routine in the NAVSIM scorer when it is first mentioned.
  5. [§3.4 and §4.1] The paper says that a second stock run 'exactly reproduces all per-token score outputs,' but it does not state whether this second run used a different machine or the same audited backend; please specify the conditions of the reproducibility run.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity: the score-basis failure is measured, and the diagnostic controls are independent of the formal definition.

full rationale

Score 0. No step in the paper's argument reduces to its own inputs by construction. The formal condition in Sec. 3.3 defines shared-failure forgiveness collapse via equations for the reference failure mask F_j and the overwritten score \tilde m_j, but that definition does not entail the empirical observations. The audit then tests the condition with independent measurements: blind-probe rankings, fail-to-pass overwrite rates, dependency-stack controls, solver substitution, and two-sided finite-difference diagnostics. The stock reversal (79.6 vs 74.0), the 99.98% DAC/DDC/lane-keeping overwrite rates, and the reversal to 13.3 vs 88.9 under the two-sided diagnostic are measurements, not fits. No parameter is fitted to the target result; the two blind probes are described with fixed parameterizations used across all tokens and score bases. No load-bearing self-citation occurs: references are to NAVSIM, LEAD, and related benchmark literature, and the audit's core evidence is the reproduced divergence on a fresh install versus a same-source control stack with identical scorer source hashes. The paper candidly scopes the solver-replacement and stack controls to the 450-token navhard stage-1 pool and the fixed 32-token diagnostic, not the full 12,146-token navtest split; that narrows the causal generalization of the abstract's 'direct trigger' statement, but it is a limitation in evidence scope, not circularity. For the circularity question, the derivation chain is self-contained and empirically grounded.

Assumptions & free parameters 0 free parameters · 4 assumptions · 0 invented entities

No numeric constants are fitted to data in this audit. The probe parameters (vdes=15, a=1.5, b=3.0, 4s horizon) and fallback thresholds (100m path, 50m step) are fixed a priori, and the authors vary fallback thresholds in a robustness check. The central claim rests on domain assumptions about the NAVSIM scorer's shared reference path, the operational definition of the affected stack, and the representativeness of the diagnostic interventions.

assumptions (4)
  • domain assumption NAVSIM's original-scene scorer re-simulates both the proposal and the logged human reference through the same BatchLQR velocity refit, and applies reference-conditioned forgiveness as in Eq. (2).
    Sec. 3.3 and Sec. 4.2 rely on this shared path; if the reference's failure mask were computed on a different basis, the forgiveness overwrite would not occur.
  • domain assumption The labeled 'affected documented-stack condition' (Python 3.9.23, NumPy 1.23.4, SciPy 1.13.1, OpenCV 4.9.0, OpenBLAS 0.3.20) is the configuration under which the headline numbers are produced.
    Sec. 3.2 defines the condition operationally; the authors do not identify a library-internal cause, only a dependency-sensitive difference.
  • domain assumption The two-sided finite-difference diagnostic fairly approximates a stable score basis; its symmetric application to proposal and reference gives the ordering used to conclude blind-last recovery.
    Sec. 4.3 uses this diagnostic as an independent basis; it is not the stock scorer, and its validity as the reference ordering is corroborated by solver and stack controls rather than independently established.
  • domain assumption The 32-token diagnostic set, despite its unavailable original sampling rule, is adequate for mechanism reproduction.
    Table 6 and Sec. 3.4 disclose that the parent pool rule is unavailable; the authors use the set only as scoped mechanism evidence, not as a prevalence sample.

how reviews work

0 comments
Cite this review

Pith. "Pith review of When Shared Rollouts Fail in Defensive Driving Evaluation: A NAVSIM Score Basis Audit." pith.science (2026). https://pith.science/paper/XRWQPOGU

@misc{pith2026260804896,
  author       = {Pith},
  title        = {Pith review of: When Shared Rollouts Fail in Defensive Driving Evaluation: A NAVSIM Score Basis Audit},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/XRWQPOGU}},
  note         = {Machine review of arXiv:2608.04896}
}
read the original abstract

Defensive driving scores are useful only when they preserve distinctions between policies that observe surrounding actors and those that do not. Re-simulation benchmarks may use reference-conditioned forgiveness, under which an agent receives credit when the logged human reference fails a compliance channel. When agent and reference share an unstable rollout transformation, this rule can propagate shared reference failures into broad compliance credit. We audit this risk in NAVSIM v2.2 original scene single-stage scoring. Under the affected documented-stack condition on the audited numerical backend, the route-blind Ignore-All probe and a route-aware actor-blind probe outrank human replay and PDM-Closed over the complete 12,146-token navtest split. A fresh installation following the public specification reproduces rollout divergence on a fixed 32-token diagnostic set. A same-source dependency stack control and an exact-input diagnostic isolate dependency-sensitive numerical behavior in the shared velocity refit. On a 450-token control pool, replacing only the solver eliminates rollout divergence and restores blind-last ordering while keeping forgiveness enabled. Thus, the numerical instability is the direct trigger. Reference-conditioned forgiveness propagates the resulting shared reference failures into compliance credit. We contribute an audit protocol requiring score basis and stack disclosure, blind probes, overwrite reporting, and rollout stability tests before using such scores for defensive driving claims.

Figures

Figures reproduced from arXiv: 2608.04896 by the authors.

Figure 1
Figure 1. Shared-failure forgiveness collapse. Under stock filter-on scoring in the affected documented-stack condition (left), the actor-blind Ignore-All, which receives no surrounding-actor input, outranks human replay with 79.6 vs. 74.0 Extended Predictive Driver Model Score (EPDMS). Shared reference failure produces near-saturated fail￾to-pass overwrite in the core Drivable Area Compliance (DAC), Driving Direction Complia… view at source ↗

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

37 extracted references · 29 canonical work pages

  1. [1]

    arXiv preprint arXiv:1606.06565 (2016)

    Amodei, D., Olah, C., Steinhardt, J., Christiano, P., Schulman, J., Mané, D.: Con- crete problems in AI safety. arXiv preprint arXiv:1606.06565 (2016)

    arXiv 2016

  2. [2]

    arXiv preprint arXiv:2106.11810 (2021)

    Caesar, H., Kabzan, J., Tan, K.S., Fong, W.K., Wolff, E., Lang, A., Fletcher, L., Beijbom, O., Omari, S.: nuPlan: A closed-loop ML-based planning benchmark for autonomous vehicles. arXiv preprint arXiv:2106.11810 (2021)

    arXiv 2021

  3. [3]

    Evaluation and program planning2(1), 67–90 (1979)

    Campbell, D.T.: Assessing the impact of planned social change. Evaluation and program planning2(1), 67–90 (1979)

    work page 1979

  4. [4]

    In: Proceedings of the 9th Conference on Robot Learning

    Cao, W., Hallgarten, M., Li, T., Dauner, D., Gu, X., Wang, C., Miron, Y., Aiello, M., Li, H., Gilitschenski, I., Ivanovic, B., Pavone, M., Geiger, A., Chitta, K.: Pseudo-simulation for autonomous driving. In: Proceedings of the 9th Conference on Robot Learning. Proceedings of Machine Learning Research, vol. 305, pp. 4709–

    work page

  5. [5]

    In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)

    Casas, S., Sadat, A., Urtasun, R.: MP3: A unified model to map, perceive, predict and plan. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR). pp. 14403–14412 (2021)

    work page 2021

  6. [6]

    In: Proceedings of the IEEE/CVF International Conference on Computer Vision (ICCV)

    Chen, D., Koltun, V., Krähenbühl, P.: Learning to drive from a world on rails. In: Proceedings of the IEEE/CVF International Conference on Computer Vision (ICCV). pp. 15590–15599 (2021)

    work page 2021

  7. [7]

    In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)

    Chen, D., Krähenbühl, P.: Learning from all vehicles. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR). pp. 17222–17231 (2022)

    work page 2022

  8. [8]

    In: Confer- ence on Robot Learning (CoRL)

    Chen, D., Zhou, B., Koltun, V., Krähenbühl, P.: Learning by cheating. In: Confer- ence on Robot Learning (CoRL). pp. 66–75. PMLR (2020)

    work page 2020

Show all 37 references
  1. [9]

    IEEE transac- tions on pattern analysis and machine intelligence (TPAMI)45(11), 12878–12895 (2023)

    Chitta, K., Prakash, A., Jaeger, B., Yu, Z., Renz, K., Geiger, A.: TransFuser: Imi- tation with transformer-based sensor fusion for autonomous driving. IEEE transac- tions on pattern analysis and machine intelligence (TPAMI)45(11), 12878–12895 (2023)

    2023

  2. [10]

    In: 2018 IEEE International Conference on Robotics and Automation (ICRA)

    Codevilla, F., Müller, M., López, A., Koltun, V., Dosovitskiy, A.: End-to-end driv- ing via conditional imitation learning. In: 2018 IEEE International Conference on Robotics and Automation (ICRA). pp. 4693–4700. IEEE (2018)

    2018

  3. [11]

    In: Conference on Robot Learning (CoRL)

    Dauner, D., Hallgarten, M., Geiger, A., Chitta, K.: Parting with misconceptions about learning-based vehicle motion planning. In: Conference on Robot Learning (CoRL). pp. 1268–1281. PMLR (2023)

    2023

  4. [12]

    Advances in Neural Information Processing Systems (NeurIPS)37, 28706–28719 (2024)

    Dauner, D., Hallgarten, M., Li, T., Weng, X., Huang, Z., Yang, Z., Li, H., Gilitschenski, I., Ivanovic, B., Pavone, M., et al.: NAVSIM: Data-driven non- reactive autonomous vehicle simulation and benchmarking. Advances in Neural Information Processing Systems (NeurIPS)37, 2870...

    2024

  5. [13]

    In: Proceedings of the 20th Conference on Uncertainty in Artificial Intelligence

    Ferns, N., Panangaden, P., Precup, D.: Metrics for finite Markov decision processes. In: Proceedings of the 20th Conference on Uncertainty in Artificial Intelligence. pp. 162–169. UAI ’04, AUAI Press, Arlington, Virginia, USA (2004)

    2004

  6. [14]

    Advances in Neural Information Processing Systems (NeurIPS)33, 5541–5552 (2020)

    Grimm, C., Barreto, A., Singh, S., Silver, D.: The value equivalence principle for model-based reinforcement learning. Advances in Neural Information Processing Systems (NeurIPS)33, 5541–5552 (2020)

    2020

  7. [15]

    arXiv preprint arXiv:2109.13916 (2021)

    Hendrycks, D., Carlini, N., Schulman, J., Steinhardt, J.: Unsolved problems in ML safety. arXiv preprint arXiv:2109.13916 (2021)

    2021 arXiv

  8. [16]

    In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)

    Hu, Y., Yang, J., Chen, L., Li, K., Sima, C., Zhu, X., Chai, S., Du, S., Lin, T., Wang, W., et al.: Planning-oriented autonomous driving. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR). pp. 17853–17862 (2023) 16 Z. Wei and M. Yu an...

    2023

  9. [17]

    Transactions on Machine Learning Research (TMLR) (2025)

    Hwang, J.J., Xu, R., Lin, H., Hung, W.C., Ji, J., Choi, K., Huang, D., He, T., Cov- ington, P., Sapp, B., et al.: EMMA: End-to-end multimodal model for autonomous driving. Transactions on Machine Learning Research (TMLR) (2025)

    2025

  10. [18]

    In: Proceedings of the IEEE/CVF International Conference on Computer Vision (ICCV)

    Jaeger, B., Chitta, K., Geiger, A.: Hidden biases of end-to-end driving models. In: Proceedings of the IEEE/CVF International Conference on Computer Vision (ICCV). pp. 8240–8249 (2023)

    2023

  11. [19]

    In: Proceedings of the IEEE/CVF International Conference on Computer Vi- sion (ICCV)

    Jiang,B.,Chen,S.,Xu,Q.,Liao,B.,Chen,J.,Zhou,H.,Zhang,Q.,Liu,W.,Huang, C., Wang, X.: VAD: Vectorized scene representation for efficient autonomous driv- ing. In: Proceedings of the IEEE/CVF International Conference on Computer Vi- sion (ICCV). pp. 8340–8350 (2023)

    2023

  12. [20]

    arXiv preprint arXiv:2406.06978 (2024)

    Li, Z., Li, K., Wang, S., Lan, S., Yu, Z., Ji, Y., Li, Z., Zhu, Z., Kautz, J., Wu, Z., et al.: Hydra-MDP: End-to-end multimodal planning with multi-target hydra- distillation. arXiv preprint arXiv:2406.06978 (2024)

    2024 arXiv

  13. [21]

    IEEE Transactions on Pattern Analysis and Machine Intelligence (TPAMI)47(3), 2020–2036 (2025)

    Li, Z., Wang, W., Li, H., Xie, E., Sima, C., Lu, T., Yu, Q., Dai, J.: BEVFormer: Learning bird’s-eye-view representation from LiDAR-camera via spatiotemporal transformers. IEEE Transactions on Pattern Analysis and Machine Intelligence (TPAMI)47(3), 2020–2036 (2025)

    2025

  14. [22]

    Li, Z., Yu, Z., Lan, S., Li, J., Kautz, J., Lu, T., Alvarez, J.M.: Is ego status all you need for open-loop end-to-end autonomous driving? In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR). pp. 14864–14873 (2024)

    2024

  15. [23]

    In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)

    Liao, B., Chen, S., Yin, H., Jiang, B., Wang, C., Yan, S., Zhang, X., Li, X., Zhang, Y., Zhang, Q., et al.: DiffusionDrive: Truncated diffusion model for end-to-end autonomous driving. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)...

    2025

  16. [24]

    arXiv preprint arXiv:1803.04585 (2018)

    Manheim, D., Garrabrant, S.: Categorizing variants of Goodhart’s law. arXiv preprint arXiv:1803.04585 (2018)

    2018 arXiv

  17. [25]

    In: Proceed- ings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) (2026)

    Nguyen, L., Fauth, M., Jaeger, B., Dauner, D., Igl, M., Geiger, A., Chitta, K.: LEAD: Minimizing learner–expert asymmetry in end-to-end driving. In: Proceed- ings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) (2026)

    2026

  18. [26]

    In: International Conference on Learning Representations (ICLR) (2022)

    Pan, A., Bhatia, K., Steinhardt, J.: The effects of reward misspecification: Map- ping and mitigating misaligned models. In: International Conference on Learning Representations (ICLR) (2022)

    2022

  19. [27]

    Philion, J., Fidler, S.: Lift, splat, shoot: Encoding images from arbitrary camera rigsbyimplicitlyunprojectingto3D.In:EuropeanConferenceonComputerVision (ECCV). pp. 194–210. Springer (2020)

    2020

  20. [28]

    In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)

    Prakash, A., Chitta, K., Geiger, A.: Multi-modal fusion transformer for end-to-end autonomous driving. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR). pp. 7077–7087 (2021)

    2021

  21. [29]

    In: Proceed- ings of the 6th Conference on Robot Learning

    Renz, K., Chitta, K., Mercea, O.B., Koepke, A.S., Akata, Z., Geiger, A.: PlanT: Explainable planning transformers via object-level representations. In: Proceed- ings of the 6th Conference on Robot Learning. Proceedings of Machine Learning Research, vol. 205, pp. 459–470. PMLR (2023)

    2023

  22. [30]

    In: Conference on Robot Learning (CoRL)

    Shao, H., Wang, L., Chen, R., Li, H., Liu, Y.: Safety-enhanced autonomous driving using interpretable sensor fusion transformer. In: Conference on Robot Learning (CoRL). pp. 726–737. PMLR (2023)

    2023

  23. [31]

    In: European Conference on Computer Vision (ECCV)

    Sima, C., Renz, K., Chitta, K., Chen, L., Zhang, H., Xie, C., Beißwenger, J., Luo, P., Geiger, A., Li, H.: DriveLM: Driving with graph visual question answering. In: European Conference on Computer Vision (ECCV). pp. 256–274. Springer (2024) NAVSIM Score Basis Audit 17

    2024

  24. [32]

    Advances in Neural Information Processing Systems (NeurIPS) 35, 9460–9471 (2022)

    Skalse, J., Howe, N., Krasheninnikov, D., Krueger, D.: Defining and characterizing reward gaming. Advances in Neural Information Processing Systems (NeurIPS) 35, 9460–9471 (2022)

    2022

  25. [33]

    arXiv preprint arXiv:2605.00066 (2026)

    Wang, Y., Jiang, A., Wang, S., Heng, Y., Yang, H., Chen, Y., Sun, H.: Do open- loop metrics predict closed-loop driving? A cross-benchmark correlation study of NAVSIM and Bench2Drive. arXiv preprint arXiv:2605.00066 (2026)

    2026 arXiv

  26. [34]

    Advances in Neural Information Processing Systems (NeurIPS)35, 6119–6132 (2022)

    Wu, P., Jia, X., Chen, L., Yan, J., Li, H., Qiao, Y.: Trajectory-guided control pre- diction for end-to-end autonomous driving: A simple yet strong baseline. Advances in Neural Information Processing Systems (NeurIPS)35, 6119–6132 (2022)

    2022

  27. [35]

    arXiv preprint arXiv:2305.10430 (2023)

    Zhai, J.T., Feng, Z., Du, J., Mao, Y., Liu, J.J., Tan, Z., Zhang, Y., Ye, X., Wang, J.: Rethinking the open-loop evaluation of end-to-end autonomous driv- ing in nuScenes. arXiv preprint arXiv:2305.10430 (2023)

    2023 arXiv

  28. [36]

    arXiv preprint arXiv:1910.03088 (2019)

    Zhan, W., Sun, L., Wang, D., Shi, H., Clausse, A., Naumann, M., Kummerle, J., Konigshof, H., Stiller, C., de La Fortelle, A., et al.: INTERACTION dataset: An international, adversarial and cooperative motion dataset in interactive driving scenarios with semantic maps. arXiv pr...

    2019 arXiv

  29. [37]

    In: International Conference on Learning Representations (ICLR) (2021)

    Zhang, A., McAllister, R., Calandra, R., Gal, Y., Levine, S.: Learning invariant representations for reinforcement learning without reconstruction. In: International Conference on Learning Representations (ICLR) (2021)

    2021

Pith tools

Reviewed August 6, 2026 · model on record in the stance chip above.