Pith. sign in

REVIEW 4 major objections 6 minor 47 references

Fractional Collisions: A Framework for Risk Estimation of Counterfactual Conflicts using Autonomous Driving Behavior Simulations

T0 review · 4 major / 6 minor · reviewed 2026-08-07 · deepseek-v4-flash

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

Pith's one-line read Counterfactual traffic conflicts can be scored as fractional collisions that sum to real crash counts within 1%.

desk verdict The fractional collision metric is worth taking seriously, but the paper's headline 1% validation is a selection artifact, not a confirmation, and the ADS comparison mixes discrete and fractional counts. read the letter →

arxiv 2506.07540 v1 pith:KQCJZFXU submitted 2025-06-09 cs.RO cs.SYeess.SY

Sreeja Roy-Singh , Sarvesh Kolekar , Daniel P. Bonny , Kyle Foss This is my paper
classification cs.ROcs.SYeess.SY
keywords counterfactualsimulationfractionalcollisionscollisionriskestimationautonomousdrivingsafetyhumanrespondermodelingbehavioraluncertaintypositivebalancecrashseveritymapping
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

This paper proposes a method for estimating the collision risk of a simulated two-agent traffic conflict without waiting for an actual crash. Instead of predicting one outcome, it models the human responder's reaction probabilistically, weights the resulting severities by their joint probability, and reports the expected loss as a fractional collision value that can be summed across scenes. The paper's central verification claim is that, aggregated over a set of reconstructed naturalistic scenes, these fractional collisions reproduce the discrete ground-truth collision count almost exactly (17.15 fractional versus 17 actual collisions). The paper argues that this aggregate agreement is what makes the metric usable for automated-driving-system safety evaluation, letting developers compare an ADS against modeled human responders and flag high-risk conflicts.

What carries the argument

The load-bearing mechanism is the fractional collision scoring pipeline. For each two-agent conflict, the framework classifies the conflict type, assigns initiator and responder roles, determines the responder's point of reaction (the timestamp at which the human could first perceive the conflict and the clock for reaction time starts), draws counterfactual trajectories from joint distributions over human reaction time, longitudinal jerk, and steady-state acceleration, and maps each trajectory's velocity differential at impact through a crash severity model to a severity level. The output is a probability-weighted sum of severities, the fractional collision, which can be added across conflicts. The property that makes this useful is the aggregate hypothesis: over an unbiased sample of mileage, the summed fractional collisions should match the discrete ground-truth collision count, so the continuous risk signal remains calibrated to real-world crash exposure.

What would settle it

Apply the framework to a large random sample of naturalistic miles, running conflict detection, point-of-reaction labeling, and the responder models exactly as specified, and compare the summed fractional collisions with the actual crash count in that same sample by severity level; a divergence beyond the claimed 1 percent on total collisions, or systematic per-severity bias, would falsify the aggregate identity. A more targeted falsifier is a controlled early-versus-late shift of the point of reaction by about 0.2 seconds, which should change the reaction-time distribution enough to move the aggregate beyond the claimed tolerance if the metric is truly calibrated.

Watch

Extended reading notes

Core claim

The central claim is that a traffic conflict's risk can be expressed as a fractional collision: the probability-weighted sum of loss severities (no collision, L2 property damage, L1 injury, L0 higher-severity injury) over the distribution of plausible human responses after the point of reaction. When these fractional values are summed over a sample of mileage that represents the human responder population, the total is claimed to approximate the number of discrete collisions that actually occurred in that sample. The paper reports this on its selected naturalistic verification set, where the total fractional collision count was 17.15 against 17 ground-truth collisions, roughly 1% higher, and it positions this aggregate property as the justification for using fractional collisions in positive-risk-balance evaluations of ADS behavior.

Load-bearing premise

The aggregate accuracy depends on the point-of-reaction heuristic and the human reaction-time, jerk, and acceleration distributions being representative of real drivers across every operational design domain and conflict type; if the reaction-clock start is biased, the fractional totals shift across the board.

Editorial extensions

If this is right

  • ADS software can be scored continuously per release, with each simulated conflict contributing an expected loss rather than a binary crash/no-crash outcome.
  • Rare but dangerous agent-initiated scenarios become findable: the paper reports that on 250k miles of replay simulation only 13 of 376 conflicts showed a modeled human performing better than the ADS, allowing those scenes to be targeted for improvement.
  • ADS-initiated collisions that never materialize on road can still be counted as fractional risk, e.g., six conflicts totaling 0.4 injury-causing and 1.7 property-damaging fractional collisions in the 250k-mile replay.
  • The naturalistic aggregate check is reusable: any developer can validate a simulator or agent model by comparing summed fractional collisions against real crash counts in an unbiased mileage sample.
  • Positive risk balance becomes a quantitative, scene-level comparison between the ADS outcome and the distribution of human outcomes rather than a single fleet-level ratio.

Reading between the lines

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

  • If the aggregate agreement holds on uncurated random mileage, fractional collisions could be promoted from a development signal to an exposure-normalized risk metric for safety-case submissions, analogous to crash rates per mile but computed from counterfactuals.
  • The same pipeline could absorb other uncertainty sources—sensor latency, controller error, responder aggressiveness—by widening the probability mass functions, a direction the paper names but does not implement.
  • A sharp calibration test would be to run the framework with two different point-of-reaction definitions (surprise-based versus omniscient) and track how the aggregate total moves; the paper acknowledges point-of-reaction sensitivity but does not quantify it.
  • Pose divergence in open-loop replays is tolerated for recall, but a safety case built on fractional collisions would eventually need a de-biasing step for that inflation.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

4 major / 6 minor

Summary. The paper introduces 'fractional collisions,' a Monte-Carlo-style risk metric that converts counterfactual two-agent conflicts into a continuous severity-weighted collision count. The framework classifies conflict type, identifies initiator and responder, estimates the responder's Point of Reaction (PoR), samples human kinematic reaction parameters (HRT, jerk, acceleration) from literature and proprietary distributions, computes delta-v via a crash severity model, and aggregates the resulting severity probabilities into a fractional collision score. The authors verify the approach on reconstructed SHRP2 and Nexar traffic scenes, reporting a 1% aggregate match against ground-truth collisions on 82 SHRP2 scenes, then apply the framework to evaluate an ADS in both synthetic and quarter-million-mile replay settings, claiming a 4x reduction in naturalistic collisions and ~62% reduction in fractional collision risk.

Significance. If the aggregate verification held, fractional collisions would be a valuable tool for ADS safety benchmarking because they are more frequent than discrete collisions, interpretable against real-world collision statistics, and usable for both AUR and PRB arguments. The manuscript is clearly written, describes a plausible end-to-end workflow, includes useful scene-level quality checks, and does not fit its parameters to the verification aggregate, which avoids a circular validation. However, the headline 'within 1%' result is not yet established, because the selected SHRP2 scene mix is not mileage-representative and the apparent agreement is a cancellation of opposite errors.

major comments (4)
  1. [§4.2, Table 1] The claimed 1% aggregate match does not test the stated aggregate hypothesis because the selected SHRP2 scenes are not proportionally representative of naturalistic driving. The manuscript states that the SHRP2 database has a collision-to-near-collision (C:NC) ratio of 1:15, yet the selected set has 17:65 (≈1:3.8), oversampling collisions by a factor of roughly four. Table 1 shows that the 17 collision scenes contribute 12.6 fractional collisions while the 65 near-collision scenes contribute 4.56. Reweighting the NC term to the naturalistic 255-per-17-collision ratio gives 12.6 + 4.56 × (255/65) ≈ 30.5 fractional collisions against 17 ground-truth collisions, a 79% overestimate. Thus the 17.15 ≈ 17 agreement is a selection artifact that cancels a 26% under-prediction on collision scenes with a 4.56-unit over-prediction on NC scenes. The paper must re-run the aggregate comparison on a mileage-representative sample (or reweight the existing one) and report the result, rather than presenting the selected-sample agreement as evidence for the framework's aggregate accuracy.
  2. [§4.2, Nexar discussion] The paper attributes Nexar's 40% under-prediction to the procured data's disproportionately high C:NC ratio, arguing that a dataset must contain statistically representative NC scenarios to predict total collisions accurately. The same logic applies symmetrically to the SHRP2 selection: the selected 17:65 ratio is also far from the naturalistic 1:15 ratio, and the paper's own reasoning implies that the SHRP2 aggregate cannot verify the unbiased-mileage hypothesis. A consistent treatment of both datasets is required—either reweight both to naturalistic conflict frequencies or present the 1% agreement only as a property of the hand-picked scene set, not as a general validation.
  3. [§3.2] The Point of Reaction (PoR) is a critical input: the manuscript notes that the human reaction time (HRT) distributions are measured relative to PoR and that the models are 'only valid when PoR is properly defined.' Yet PoR is currently determined by heuristics with human QA rather than an independently validated detector. Because any systematic bias in PoR timing would shift all HRT distributions and change every fractional collision estimate, the framework's accuracy claims---including the aggregate comparison in §4.2---depend on an unvalidated component. Please provide a sensitivity analysis (e.g., perturbing PoR by ±100–300 ms and reporting the resulting change in aggregate fractional collisions) or compare the heuristic PoR labels against an established surprise-based framework such as Ref. [12].
  4. [§4.1] The scene-level checks described in §4.1 are presented as verification, but two of the three checks only confirm internal consistency (reconstruction matches GT, and the no-reaction model is no better than GT). The more meaningful statistic---that the most probable severity matches GT in 91% of SHRP2 scenes---is reported without confidence intervals or a comparison to a baseline classifier, making it difficult to judge whether the model is genuinely predictive or merely reproducing the dominant severity class. Reporting a baseline (e.g., always-predict-NC accuracy) and per-conflict-type accuracy would strengthen the scene-level evidence.
minor comments (6)
  1. [§3.3] Typo: 'V olvo' should be 'Volvo'; the reference list also uses 'IS0 26262' instead of 'ISO 26262' in the text.
  2. [§3.4] In the paragraph describing the probabilistic collision value, the last term is listed as 'P(L2)' twice; one of these should presumably be 'P(L0)' to match the four-severity taxonomy.
  3. [§4.2] The sentence 'the number of discrete collisions that a NRM would had in those scenes' contains a grammatical error ('would had' should be 'would have had').
  4. [§5.1] 'The PRB gap in 3.5% scenes' should be 'in 3.5% of scenes'.
  5. [§5.2] The six ADS-initiated conflicts are described as surfacing from a quarter-million-mile replay, but the detection criterion is not specified beyond 'non-zero fractional collisions.' Clarify whether these are all conflicts found by the SSM pipeline and how false positives are handled.
  6. [References] Reference [18] is cited as 'ISO/PAS 26262:2018(en)' but ISO 26262 is an international standard (ISO 26262:2018) rather than a PAS; the citation should be corrected to avoid confusion.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: fractional collisions are computed from externally sourced behavioral distributions and are not fitted to the ground-truth aggregate; the 1% match is a reported verification result, not an identity.

full rationale

The paper's derivation chain is not circular. Fractional collisions are probability-weighted severity sums (Section 3.4) built from responder behavior distributions (HRT, jerk, acceleration) attributed to external SHRP2-based studies, laboratory/field studies, and proprietary Nuro logs; the paper does not fit these distributions to the 17/65 validation scenes or to the GT total of 17. The aggregate comparison in Section 4.2 (17.15 vs 17) is reported as a verification result, not imposed by an equation or by a fitted parameter. The paper candidly notes that the selected SHRP2 C:NC ratio (17:65) differs from the database's 1:15 and states that 'the dataset must contain statistically representative NC scenarios' for accurate totals; that caveat concerns sample representativeness rather than circularity. The only author self-citation ([22], a Nature Communications risk-based driver model) is background motivation and is not load-bearing for the fractional-collision derivation. The 1% agreement may be fragile under reweighting, as a sampling critique, but that fragility is a validity limitation, not a reduction of the prediction to its inputs.

Assumptions & free parameters 3 free parameters · 4 assumptions · 0 invented entities

The fractional collision score depends on a set of chosen thresholds (6 mph, 20 mph, 5/15 mph for VRUs), on HRT and jerk/acceleration distributions from literature and proprietary logs, and on the assumption that these distributions represent the human population. The aggregate verification hypothesis is an added assumption, not a derived theorem. No new physical entities are introduced, but 'fractional collision' is a new aggregate metric.

free parameters (3)
  • velocity-delta severity thresholds = 6 mph (L1/L2), 20 mph (L0); VRU thresholds 5 and 15 mph
    Chosen in Section 3.3 to approximate VTTI L1/L2 and add an L0 tier. These thresholds directly determine the severity class probabilities in fractional collisions and are not derived from data.
  • Human reaction time (HRT) distributions per conflict type = not fully specified in paper
    Core input to the fractional collision computation; taken from literature (e.g., Markkula et al. SHRP2 rear-end, Dinakar et al. cut-in) and proprietary Nuro logs. The exact distributions are not published, so the central result depends on unstated fitted values.
  • Longitudinal jerk and steady-state acceleration distributions = not specified
    Secondary kinematic parameters in the responder model (Section 3.2), sourced from literature and internal data; exact values omitted.
assumptions (4)
  • domain assumption Responders behave according to NIEON (Non-Impaired and Eyes On the conflict) assumptions
    Section 3.2 assumes the responder has intent and ability to react; inattentive agents are only separately modeled via population fractions. This bounds the behavior model to attentive drivers.
  • ad hoc to paper Aggregate fractional collisions should equal ground truth collisions on unbiased mileage
    Section 4 states this as the hypothesis for verification; it is an assumed relationship between a probabilistic model sum and discrete observed counts, not proven. The paper's own Nexar result (40%) shows the assumption fails on biased samples.
  • domain assumption A few kinematic parameters (HRT, jerk, acceleration) suffice to model human evasive trajectories
    Section 3.2 models responder behavior with only these parameters; lateral maneuvers are treated as a QA step rather than modeled, and Section 6 concedes cyclist nudges are not captured.
  • standard math Momentum-based crash severity model from Ref [6] is valid for the simulated impacts
    Section 3.3 uses a published momentum-based contact algorithm for vehicle deltav; treated as trusted prior work.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Fractional Collisions: A Framework for Risk Estimation of Counterfactual Conflicts using Autonomous Driving Behavior Simulations." pith.science (2026). https://pith.science/paper/KQCJZFXU

@misc{pith2026250607540,
  author       = {Pith},
  title        = {Pith review of: Fractional Collisions: A Framework for Risk Estimation of Counterfactual Conflicts using Autonomous Driving Behavior Simulations},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/KQCJZFXU}},
  note         = {Machine review of arXiv:2506.07540}
}
read the original abstract

We present a methodology for estimating collision risk from counterfactual simulated scenarios built on sensor data from automated driving systems (ADS) or naturalistic driving databases. Two-agent conflicts are assessed by detecting and classifying conflict type, identifying the agents' roles (initiator or responder), identifying the point of reaction of the responder, and modeling their human behavioral expectations as probabilistic counterfactual trajectories. The states are used to compute velocity differentials at collision, which when combined with crash models, estimates severity of loss in terms of probabilistic injury or property damage, henceforth called fractional collisions. The probabilistic models may also be extended to include other uncertainties associated with the simulation, features, and agents. We verify the effectiveness of the methodology in a synthetic simulation environment using reconstructed trajectories from 300+ collision and near-collision scenes sourced from VTTI's SHRP2 database and Nexar dashboard camera data. Our methodology predicted fractional collisions within 1% of ground truth collisions. We then evaluate agent-initiated collision risk of an arbitrary ADS software release by replacing the naturalistic responder in these synthetic reconstructions with an ADS simulator and comparing the outcome to human-response outcomes. Our ADS reduced naturalistic collisions by 4x and fractional collision risk by ~62%. The framework's utility is also demonstrated on 250k miles of proprietary, open-loop sensor data collected on ADS test vehicles, re-simulated with an arbitrary ADS software release. The ADS initiated conflicts that caused 0.4 injury-causing and 1.7 property-damaging fractional collisions, and the ADS improved collision risk in 96% of the agent-initiated conflicts.

Figures

Figures reproduced from arXiv: 2506.07540 by the authors.

Figure 1
Figure 1. Simulation overview in a 2-body conflict. Agent C initi [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 2
Figure 2. Summary of fractional collision workflow. Inset shows [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗
Figure 3
Figure 3. Example of an HRT distribution for a human respon [PITH_FULL_IMAGE:figures/full_fig_p004_3.png] view at source ↗
Figures from the paper (2 more)
Figure 4
Figure 4. Figure 4: Example of an agent-initiated conflict with non-zero [PITH_FULL_IMAGE:figures/full_fig_p007_4.png]
Figure 5
Figure 5. Figure 5: [Left] Example of an ADS-initiated conflict with non-zero fractional collision risk in simulation. The ADS unnecessarily yields [PITH_FULL_IMAGE:figures/full_fig_p008_5.png]

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

47 extracted references · 42 canonical work pages

  1. [12]

    Modeling road user response timing in naturalistic traffic conflicts: A surprise-based framework

    Johan Engstr ¨om, Shu-Yuan Liu, Azadeh Dinparastdjadid, and Camelia Simoiu. Modeling road user response timing in naturalistic traffic conflicts: A surprise-based framework. Accident Analysis and Prevention, 198:107460, 2024. 3

    work page 2024

  2. [1]

    Active road user interactions with autonomous vehicles: Proactive safety as- sessment

    Abdul Razak Alozi and Mohamed Hussein. Active road user interactions with autonomous vehicles: Proactive safety as- sessment. Transportation research record, 2677(7):74–89,

    work page

  3. [2]

    Safety assessment of autonomous cars using verification tech- niques

    Matthias Althoff, Olaf Stursberg, and Martin Buss. Safety assessment of autonomous cars using verification tech- niques. In 2007 American Control Conference, pages 4154– 4159, 2007. 2

    work page 2007

  4. [3]

    Proba- bilistic collision estimation system for autonomous vehicles

    Stefan Annell, Alexander Gratne, and Lars Svensson. Proba- bilistic collision estimation system for autonomous vehicles. In Proceedings of the IEEE 19th International Conference on Intelligent Transportation Systems (ITSC), 2016. 2

    work page 2016

  5. [4]

    Methodological challenges of scenario genera- tion validation: a rear-end crash-causation model for virtual safety assessment

    Jonas B ¨argman, Malin Sv ¨ard, Simon Lundell, and Erik Hartelius. Methodological challenges of scenario genera- tion validation: a rear-end crash-causation model for virtual safety assessment. Transportation Research Part F: Traffic Psychology and Behaviour, 104:374–410, 2024. 2

    work page 2024

  6. [5]

    The economic and societal impact of motor vehicle crashes, 2019

    Lawrence Blincoe, Ted R Miller, Jing-Shiarn Wang, David Swedler, Tom Coughlin, Bruce Lawrence, Feng Guo, Sheila Klauer, and Thomas Dingus. The economic and societal impact of motor vehicle crashes, 2019. Technical Report DOT-HS-813403, National Highway Traffic Safety Admin- istration, 2022. 5

    work page 2019

  7. [6]

    Residual crush energy partitioning, normal and tangential en- ergy losses

    Raymond M Brach, Kevin J Welsh, and R Matthew Brach. Residual crush energy partitioning, normal and tangential en- ergy losses. Crash Reconstruction Research, 138:45, 2008. 4

    work page 2008

  8. [7]

    An evaluation of bicycle- specific agility and reaction times in mountain bikers and road cyclists

    Kim Buchholtz and TL Burgess. An evaluation of bicycle- specific agility and reaction times in mountain bikers and road cyclists. South African journal of sports medicine , 32 (1):1–5, 2020. 4

    work page 2020

Show all 47 references
  1. [8]

    Counterfactual simulations applied to shrp2 crashes: The ef- fect of driver behavior models on safety benefit estimations of intelligent safety systems

    Jonas B ¨argman, Christian-Nils Boda, and Marco Dozza. Counterfactual simulations applied to shrp2 crashes: The ef- fect of driver behavior models on safety benefit estimations of intelligent safety systems. Accident Analysis and Preven- tion, 102:165–180, 2017. 4

    2017

  2. [9]

    nuscenes: A multi- modal dataset for autonomous driving

    Holger Caesar, Varun Bankiti, Alex H Lang, Sourabh V ora, Venice Erin Liong, Qiang Xu, Anush Krishnan, Yu Pan, Gi- ancarlo Baldan, and Oscar Beijbom. nuscenes: A multi- modal dataset for autonomous driving. In Proceedings of the IEEE/CVF conference on computer vision and patte...

    2020

  3. [10]

    Cummings and Ben Bauchwitz

    M.L. Cummings and Ben Bauchwitz. Identifying research gaps through self-driving car data analysis. IEEE Transac- tions on Intelligent Vehicles, pages 1–10, 2024. 1, 2

    2024

  4. [11]

    Muttart, Darlene E

    Swaroop Dinakar, Jeffrey W. Muttart, Darlene E. Edewaard, Michael Giannone, and Connor Dickson. Driver response time in cut-off scenarios from the second strategic highway research program naturalistic database. Transportation Re- search Record, 2676(2):706–717, 2022. 3, 4

    2022

  5. [13]

    Description of the shrp 2 naturalistic database and the crash, near-crash, and baseline data sets

    Jonathan M Hankey, Miguel A Perez, and Julie A McClaf- ferty. Description of the shrp 2 naturalistic database and the crash, near-crash, and baseline data sets. Technical report, Virginia Tech Transportation Institute, 2016. 5

    2016

  6. [14]

    A new com- putational perceived risk model for automated vehicles based on potential collision avoidance difficulty (pcad)

    Xiaolin He, Riender Happee, and Meng Wang. A new com- putational perceived risk model for automated vehicles based on potential collision avoidance difficulty (pcad). Trans- portation Research Part C, 166(104751), 2024. 2

    2024

  7. [15]

    Gaia-1: A generative world model for au- tonomous driving

    Anthony Hu, Lloyd Russell, Hudson Yeo, Zak Murez, George Fedoseev, Alex Kendall, Jamie Shotton, and Gian- luca Corrado. Gaia-1: A generative world model for au- tonomous driving. arXiv preprint arXiv:2309.17080, 2023. 2

    2023 arXiv

  8. [16]

    Emergency braking performance of motorcycle riders: skill identification in a real-life perception-action task designed for training purposes

    Pedro Huertas-Leyva, Marilee Nugent, Giovanni Savino, Marco Pierini, Niccol `o Baldanzini, and Simon Rosalie. Emergency braking performance of motorcycle riders: skill identification in a real-life perception-action task designed for training purposes. Transportation research ...

    2019

  9. [17]

    Road vehicles — safety of the intended functionality

    ISO/PAS 21448:2022(en). Road vehicles — safety of the intended functionality. Standard, International Organization for Standardization, Geneva, CH, 2022. 1

    2022

  10. [18]

    Road vehicles — functional safety

    ISO/PAS 26262:2018(en). Road vehicles — functional safety. Standard, International Organization for Standard- ization, Geneva, CH, 2018. 4

    2018

  11. [19]

    Road vehicles — safety and arti- ficial intelligence

    ISO/PAS 8800:2024(en). Road vehicles — safety and arti- ficial intelligence. Standard, International Organization for Standardization, Geneva, CH, 2024. 1

    2024

  12. [20]

    Nidhi Kalra and Susan M. Paddock. Driving to safety: How many miles of driving would it take to demonstrate au- tonomous vehicle reliability? Transportation Research Part A: Policy and Practice, 94:182–193, 2016. 1

    2016

  13. [21]

    Evaluation of autonomous driving safety by operational de- sign domains (odd) in mixed traffic

    Hoseon Kim, Jieun Ko, Cheol Oh, and Seoungbum Kim. Evaluation of autonomous driving safety by operational de- sign domains (odd) in mixed traffic. Sustainability, 16(22): 9672, 2024. 3

    2024

  14. [22]

    Human-like driving behaviour emerges from a risk-based driver model

    Sarvesh Kolekar, Joost De Winter, and David Abbink. Human-like driving behaviour emerges from a risk-based driver model. Nature communications, 11(1):1–13, 2020. 2

    2020

  15. [23]

    Breaking the tyranny of net risk metrics for automated vehicle safety

    Philip Koopman, William H Widen, et al. Breaking the tyranny of net risk metrics for automated vehicle safety. Safety-Critical Systems eJournal, 3(1), 2024. 1

    2024

  16. [24]

    Kusano, Kurt Beatty, Scott Schnelle, Francesca Favaro, Cam Crary, and Trent Victor

    Kristofer D. Kusano, Kurt Beatty, Scott Schnelle, Francesca Favaro, Cam Crary, and Trent Victor. Collision avoidance testing of the waymo automated driving system, 2022. 4

    2022

  17. [25]

    Framework for a conflict typology including contributing factors for use in ads safety evaluation

    Kristofer D Kusano, John M Scanlon, Mattias Br ¨annstr¨om, Johan Engstr¨om, and Trent Victor. Framework for a conflict typology including contributing factors for use in ads safety evaluation. In 27th International Technical Conference on the Enhanced Safety of Vehicles (ESV) ...

    2023

  18. [26]

    Modular vehicle sensing, assisting con- nected system, 2022

    Lev Yitzhak Lavy. Modular vehicle sensing, assisting con- nected system, 2022. US Patent 11,405,235. 5

    2022

  19. [27]

    Formal validation of probabilistic collision risk estimation for autonomous driv- ing

    Philippe Ledent, Anshul Paigwar, Alessandro Renzaglia, Radu Mateescu, and Christian Laugier. Formal validation of probabilistic collision risk estimation for autonomous driv- ing. In 2019 IEEE International Conference on Cybernet- ics and Intelligent Systems (CIS) and IEEE Con...

    2019

  20. [28]

    A method for predicting crash configurations using counterfac- tual simulations and real-world data

    Alexandros Leledakis, Magdalena Lindman, Jonas ¨Osth, Li- nus W ˚agstr¨om, Johan Davidsson, and Lotta Jakobsson. A method for predicting crash configurations using counterfac- tual simulations and real-world data. Accident Analysis and Prevention, 150:105932, 2021. 2, 3

    2021

  21. [29]

    Learning autonomous vehicle safety concepts from demonstrations

    Karen Leung, Sushant Veer, Edward Schmerling, and Marco Pavone. Learning autonomous vehicle safety concepts from demonstrations. In 2023 American Control Conference (ACC), pages 3193–3200. IEEE, 2023. 2

    2023

  22. [30]

    A farewell to brake reaction times? kinematics-dependent brake response in naturalistic rear-end emergencies

    Gustav Markkula, Johan Engstr ¨om, Johan Lodin, Jonas B¨argman, and Trent Victor. A farewell to brake reaction times? kinematics-dependent brake response in naturalistic rear-end emergencies. Accident Analysis and Prevention, 95: 209–226, 2016. 4

    2016

  23. [31]

    Fuzzy surro- gate safety metrics for real-time assessment of rear-end colli- sion risk

    Konstantinos Mattas, Michail Makridis, George Botzoris, Akos Kriston, Fabrizio Minarini, Basil Papadopoulos, Fab- rizio Re, Greger Rognelund, and Biagio Ciuffo. Fuzzy surro- gate safety metrics for real-time assessment of rear-end colli- sion risk. a study based on empirical o...

    2020

  24. [32]

    Assessing and explaining collision risk in dynamic environments for autonomous driving safety

    Richa Nahata, Daniel Omeiza, Rhys Howard, and Lars Kunze. Assessing and explaining collision risk in dynamic environments for autonomous driving safety. In 2021 IEEE international intelligent transportation systems conference (ITSC), pages 223–230. IEEE, 2021. 2

    2021

  25. [33]

    Val- idation of human benchmark models for automated driving system approval: How competent and careful are they really? Accident Analysis and Prevention, 213:107922, 2025

    Pierluigi Olleja, Gustav Markkula, and Jonas B ¨argman. Val- idation of human benchmark models for automated driving system approval: How competent and careful are they really? Accident Analysis and Prevention, 213:107922, 2025. 2

    2025

  26. [34]

    Passenger car safety beyond adas: defining remain- ing accident configurations as future priorities

    Martin ¨Ostling, Nils Lubbe, Hanna Jeppsson, and Pradeep Puthan. Passenger car safety beyond adas: defining remain- ing accident configurations as future priorities. In The 26th International Technical Conference on the Enhanced Safety of Vehicles, Eindhoven, Netherlands, page...

    2019

  27. [35]

    Modeling driver risk perception on city roads using deep learning

    Peng Ping, Yuan Sheng, Wenhu Qin, Chiyomi Miyajima, and Kazuya Takeda. Modeling driver risk perception on city roads using deep learning. IEEE Access , 6:68850–68866,

  28. [36]

    Scanlon, Kristofer D

    John M. Scanlon, Kristofer D. Kusano, Tom Daniel, Christo- pher Alderson, Alexander Ogle, and Trent Victor. Waymo simulated driving behavior in reconstructed fatal crashes within an autonomous vehicle operating domain. Accident Analysis and Prevention, 163:106454, 2021. 2, 3

    2021

  29. [37]

    Waymo public road safety performance data

    Matthew Schwall, Tom Daniel, Trent Victor, Francesca Favaro, and Henning Hohnhold. Waymo public road safety performance data. arXiv preprint arXiv:2011.00038, 2020. 1

    2011 arXiv

  30. [38]

    On a formal model of safe and scalable self-driving cars

    Shai Shalev-Shwartz, Shaked Shammah, and Amnon Shashua. On a formal model of safe and scalable self-driving cars. arXiv preprint arXiv:1708.06374, 2017. 2

    2017 arXiv

  31. [39]

    Wijayaratna, Navreet Virdi, and Vinayak Dixit

    Amolika Sinha, Sai Chand, Kasun P. Wijayaratna, Navreet Virdi, and Vinayak Dixit. Comprehensive safety assessment in mixed fleets with connected and automated vehicles: A crash severity and rate evaluation of conventional vehicles. Accident Analysis and Prevention, 142:105567, 2020. 3

    2020

  32. [40]

    Risk control is not risk adjustment: the zero-risk theory of driver behaviour and its implications

    Heikki Summala. Risk control is not risk adjustment: the zero-risk theory of driver behaviour and its implications. Er- gonomics, 31(4):491–506, 1988. 2

    1988

  33. [41]

    Integrated safety: Es- tablishing links for a comprehensive virtual tool chain

    Linus W ˚agstr¨om, Alexandros Leledakis, Jonas ¨Osth, Mag- dalena Lindman, and Lotta Jakobsson. Integrated safety: Es- tablishing links for a comprehensive virtual tool chain. In 26th International Technical Conference on the Enhanced Safety of Vehicles (ESV): Technology: Enab...

    2019

  34. [42]

    Safety of autonomous vehicles.Journal of advanced transportation, 2020(1):8867757, 2020

    Jun Wang, Li Zhang, Yanjun Huang, and Jian Zhao. Safety of autonomous vehicles.Journal of advanced transportation, 2020(1):8867757, 2020. 1

    2020

  35. [43]

    Criticality metrics for automated driving: A review and suitability analysis of the state of the art

    Lukas Westhofen, Christian Neurohr, Tjark Koopmann, Martin Butz, Barbara Sch ¨utt, Fabian Utesch, Birte Neurohr, Christian Gutenkunst, and Eckard B ¨ode. Criticality metrics for automated driving: A review and suitability analysis of the state of the art. Archives of Computati...

    2023

  36. [44]

    Harmonized approaches for baseline creation in prospective safety performance assess- ment of driving automation systems

    Peter Wimmer, O Op Den Camp, Hendrik Weber, Henri Chajmowicz, Michael Wagner, Jorge Lorente Mallada, Felix Fahrenkrog, and Florian Denk. Harmonized approaches for baseline creation in prospective safety performance assess- ment of driving automation systems. In 27th internatio...

    2023

  37. [45]

    Exploring the in- fluence of simulated road environments on cyclist behavior

    Jing Xu, Yingzi Lin, and David Schmidt. Exploring the in- fluence of simulated road environments on cyclist behavior. International Journal of Virtual Reality, 17(3):15–26, 2017. 4

    2017

  38. [46]

    Unisim: A neural closed-loop sensor simulator

    Ze Yang, Yun Chen, Jingkang Wang, Sivabalan Mani- vasagam, Wei-Chiu Ma, Anqi Joyce Yang, and Raquel Ur- tasun. Unisim: A neural closed-loop sensor simulator. In Proceedings of the IEEE/CVF Conference on Computer Vi- sion and Pattern Recognition, pages 1389–1399, 2023. 2

    2023

  39. [47]

    How would autonomous vehicles behave in real-world crash scenarios? Accident Analysis and Prevention , 202: 107572, 2024

    Rui Zhou, Guoqing Zhang, Helai Huang, Zhiyuan Wei, Hanchu Zhou, Jieling Jin, Fangrong Chang, and Jiguang Chen. How would autonomous vehicles behave in real-world crash scenarios? Accident Analysis and Prevention , 202: 107572, 2024. 2 10

    2024

Pith tools

Reviewed August 7, 2026 · model on record in the stance chip above.