pith. sign in

arxiv: 2605.13125 · v1 · pith:2VL7QC6Tnew · submitted 2026-05-13 · 💻 cs.RO

MoCCA: A Movable Circle Probability of Collision Approximation

Tobias Kern , Christian Birkner This is my paper

Pith reviewed 2026-05-14 18:20 UTC · model grok-4.3

classification 💻 cs.RO
keywords probability of collisioncircle approximationautonomous drivingvehicle safetyerror boundsafety marginreal-time estimationsensor uncertainty
0
0 comments X

The pith

Optimizing the position of one circle per vehicle cuts over-conservatism in collision probability while matching the speed of fixed single-circle methods.

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

Automated driving systems must estimate the probability that two vehicles will collide despite sensor noise and uncertain orientations. Fixed single-circle bounds are fast but often overestimate risk, while multi-circle bounds are more accurate yet too slow for real time. MoCCA places one movable circle on each vehicle and shifts its center to minimize the distance between the circles. This keeps the calculation as cheap as ordinary single-circle methods yet lowers the excess conservatism. The paper derives an explicit upper bound on the remaining approximation error that depends mainly on how far apart the vehicles are and how much their orientations vary, then shows a safety margin that can be set using only the orientation variance to cover partial sensor coverage cases.

Core claim

By moving the center of a single circle on each vehicle so that the distance between the two circles is minimized, the approximation achieves the computational cost of a standard single-circle method while reducing over-conservatism in the probability of collision. An upper bound on the approximation error is proven to depend primarily on inter-vehicle distance and orientation variance. A safety distance margin that compensates for possible underestimation under partial coverage can be calibrated from orientation variance alone.

What carries the argument

The movable circle on each vehicle whose center is shifted to minimize the relative distance to the opposing circle.

If this is right

  • The method runs at the same computational cost as ordinary single-circle approximations.
  • The error remains bounded by a quantity that grows with inter-vehicle distance and orientation variance.
  • A safety margin can be added using only the measured orientation variance, without needing full shape details.
  • The bound covers the partial-coverage cases that arise when sensors see only part of another vehicle.

Where Pith is reading between the lines

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

  • The same movable-circle idea could be applied to other convex bounding shapes to test whether the speed-accuracy trade-off improves further.
  • In dense traffic the lower conservatism might permit tighter but still safe trajectories that fixed-circle methods would block.
  • The variance-only margin could be updated continuously from a vehicle's own tracking filter without extra external data.

Load-bearing premise

The optimization step that chooses the circle centers truly minimizes relative distance without creating fresh underestimation errors larger than the derived bound, and that bound remains valid for the partial-coverage situations typical of real sensor data.

What would settle it

Run Monte Carlo sampling on a set of vehicle pairs with measured inter-vehicle distances and known orientation variances, then check whether the difference between MoCCA probability and the Monte Carlo value ever exceeds the paper's stated upper bound.

Figures

Figures reproduced from arXiv: 2605.13125 by Christian Birkner, Tobias Kern.

Figure 2
Figure 2. Figure 2: Depiction of determining the underapproximation error of our [PITH_FULL_IMAGE:figures/full_fig_p002_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: Scenario A) testcase 1: Head-on colliding vehicles at 100 % overlap; testcase 2: Head-on colliding vehicles at 50 % overlap; testcase 3: Ve passing Vo at a y-distance of 1.1 m; Scenario B) testcase 1: Vo crosses Ve at an x-distance of 1.1 m; testcase 2: Vo colliding sideways with the center of Ve; testcase 3: Ve and Vo approach the crossing at the same speed from the same distance. PDF has to be decorrelat… view at source ↗
Figure 5
Figure 5. Figure 5: Collision probabilities for testcase 3 of Scenario A and testcase 1 [PITH_FULL_IMAGE:figures/full_fig_p005_5.png] view at source ↗
Figure 4
Figure 4. Figure 4: Collision probabilities for each scenario (left: Scenario A; right: [PITH_FULL_IMAGE:figures/full_fig_p005_4.png] view at source ↗
Figure 6
Figure 6. Figure 6: Calculation times for MoCCA across different integration substep [PITH_FULL_IMAGE:figures/full_fig_p006_6.png] view at source ↗
read the original abstract

In automated driving, crash mitigation is crucial to ensure passenger safety. Accurate avoidance requires precise knowledge of the object's position and orientation. However, sensor noise and occlusions often result in tracking and prediction uncertainties. To account for these uncertainties, estimating the Probability of Collision (POC) is a critical requirement. While Monte Carlo sampling is a common estimation technique, its high computational demand and stochastic nature often render it unsuitable for real-time applications. Analytical POC calculations are simplified by approximating vehicle geometries using circular bounds. While multi-circle approximations offer higher fidelity than a single circumscribed circle, they significantly increase computational complexity. This paper proposes a shape approximation algorithm, MoCCA, which utilizes a single circle for each vehicle, optimized to minimize the relative distance between them. MoCCA maintains a computational efficiency comparable to standard single-circle techniques while reducing over-conservatism. To address the potential underestimation of POC inherent in partial coverage, we establish an upper bound for the approximation error, demonstrating that it depends primarily on inter-vehicle distance and orientation variance. Furthermore, we introduce a safety distance margin that can be calibrated solely based on orientation variance.

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

2 major / 1 minor

Summary. The paper proposes MoCCA, a movable-circle approximation for vehicle geometry in probability-of-collision (POC) estimation for automated driving. It replaces multi-circle or Monte Carlo methods with a single circle per vehicle whose center is optimized to minimize inter-vehicle relative distance, claiming computational cost comparable to fixed single-circle bounds while reducing over-conservatism. The manuscript asserts an upper bound on approximation error that depends primarily on inter-vehicle distance and orientation variance, together with a safety margin calibrated solely from orientation variance to compensate for partial sensor coverage.

Significance. If the error-bound derivation and the guarantee that movable-circle repositioning never exceeds the stated bound under partial coverage both hold, the method would supply a practical, real-time alternative to higher-fidelity approximations without sacrificing safety margins. The dependence of the bound and margin on only two observable quantities (distance and variance) would be a notable simplification for deployment. The absence of any derivation steps, validation data, or comparison metrics in the abstract, however, prevents a conclusive assessment of whether these advantages are realized.

major comments (2)
  1. [Abstract] Abstract: the claim that an upper bound on approximation error 'depends primarily on inter-vehicle distance and orientation variance' is asserted without any derivation, equation, or proof strategy showing that the chosen circle center remains a global minimizer of true POC (or even of relative distance) when sensor footprints cover only part of the vehicle contour.
  2. [Abstract] Abstract: the safety-distance margin 'calibrated solely based on orientation variance' presupposes that the movable-circle optimization introduces no additional under-estimation beyond the stated bound; the manuscript supplies no explicit argument that the optimization preserves this property for the partial-coverage cases typical of real sensor data.
minor comments (1)
  1. [Abstract] Abstract: the abstract refers to 'we establish an upper bound' and 'we introduce a safety distance margin' but never states the functional form of either quantity, making it impossible to verify the claimed dependence on distance and variance.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for their insightful comments on the abstract. We have revised the abstract to better convey the key aspects of the error bound and safety margin, including references to the relevant sections in the manuscript where the derivations are provided. Below, we address each major comment in detail.

read point-by-point responses

  1. Referee: [Abstract] Abstract: the claim that an upper bound on approximation error 'depends primarily on inter-vehicle distance and orientation variance' is asserted without any derivation, equation, or proof strategy showing that the chosen circle center remains a global minimizer of true POC (or even of relative distance) when sensor footprints cover only part of the vehicle contour.

    Authors: The derivation of the upper bound on the approximation error is presented in Section 3 of the manuscript. We show that by optimizing the circle center to minimize the relative distance between vehicles, the error remains bounded by a quantity that depends primarily on the inter-vehicle distance and the orientation variance. The proof involves considering the worst-case orientation uncertainty and demonstrating that the movable circle does not exceed this bound even under partial sensor coverage. To address the concern, we have revised the abstract to explicitly reference this section and briefly outline the dependence. revision: yes

  2. Referee: [Abstract] Abstract: the safety-distance margin 'calibrated solely based on orientation variance' presupposes that the movable-circle optimization introduces no additional under-estimation beyond the stated bound; the manuscript supplies no explicit argument that the optimization preserves this property for the partial-coverage cases typical of real sensor data.

    Authors: In Section 4, we provide the explicit argument that the optimization of the movable circle preserves the upper bound property for partial coverage scenarios. Specifically, we calibrate the safety margin using only the orientation variance because the geometric analysis shows that any potential under-estimation from repositioning is already accounted for within the error bound derived from distance and variance. We have updated the abstract to include a reference to this section and strengthened the discussion in the main text to clarify this point. revision: yes

Circularity Check

0 steps flagged

No circularity: error bound and margin derived from distance/variance without reducing to fitted inputs or self-citations

full rationale

The abstract and reader's summary present the upper bound on approximation error as depending on inter-vehicle distance and orientation variance, with the safety margin calibrated solely on orientation variance. No equations or steps are shown that define the bound or margin in terms of the movable-circle optimization itself, nor do they rename a fit as a prediction or import uniqueness via self-citation. The optimization minimizes relative distance as an independent step; the bound is asserted to hold afterward without tautological reduction. This matches the default expectation of a self-contained derivation.

Axiom & Free-Parameter Ledger

0 free parameters · 1 axioms · 0 invented entities

The approach rests on the domain assumption that circular bounds suffice for collision geometry and that orientation variance can be treated as an independent input for margin calibration; no free parameters are explicitly fitted in the abstract description.

axioms (1)
  • domain assumption Vehicle shapes can be conservatively bounded by circles for probability-of-collision calculations
    Invoked throughout the abstract as the basis for both single- and multi-circle methods.

pith-pipeline@v0.9.0 · 5485 in / 1177 out tokens · 49237 ms · 2026-05-14T18:20:31.165606+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

17 extracted references · 17 canonical work pages

  1. [1]

    Model-Based Probabilistic Collision Detection in Autonomous Driving,

    M. Althoff, O. Stursberg, and M. Buss, “Model-Based Probabilistic Collision Detection in Autonomous Driving,”IEEE Transactions on Intelligent Transportation Systems, vol. 10, pp. 299–310, June 2009

    work page 2009

  2. [2]

    An Integrated Approach to Maneuver-Based Trajectory Prediction and Criticality Assessment in Arbitrary Road Environments,

    M. Schreier, V . Willert, and J. Adamy, “An Integrated Approach to Maneuver-Based Trajectory Prediction and Criticality Assessment in Arbitrary Road Environments,”IEEE Transactions on Intelligent Transportation Systems, vol. 17, pp. 2751–2766, Oct. 2016

    work page 2016

  3. [3]

    Safe Nonlinear Trajectory Generation for Parallel Autonomy With a Dynamic Vehicle Model,

    W. Schwarting, J. Alonso-Mora, L. Paull, S. Karaman, and D. Rus, “Safe Nonlinear Trajectory Generation for Parallel Autonomy With a Dynamic Vehicle Model,”IEEE Transactions on Intelligent Trans- portation Systems, vol. 19, pp. 2994–3008, Sept. 2018

    work page 2018

  4. [4]

    Probabilistic constraint tighten- ing techniques for trajectory planning with predictive control,

    N. Goulet, Q. Wang, and B. Ayalew, “Probabilistic constraint tighten- ing techniques for trajectory planning with predictive control,”Journal of the Franklin Institute, vol. 359, pp. 6142–6172, Aug. 2022

    work page 2022

  5. [5]

    Probability of Collision Error Analysis,

    K. T. Alfriend, M. R. Akella, J. Frisbee, J. L. Foster, D.-J. Lee, and M. Wilkins, “Probability of Collision Error Analysis,”Space Debris, vol. 1, pp. 21–35, Mar. 1999

    work page 1999

  6. [6]

    Collision Probability Analyses For Earth-Orbiting Satellites,

    C. Ken, “Collision Probability Analyses For Earth-Orbiting Satellites,” in2001 Flight Mechanics Symposium, Proceedings of a conference sponsored NASA Goddard Space Flight Center, pp. 53–67, National Aeronautics and Space Administration, June 2001

    work page 2001

  7. [7]

    A Parametric Analysis of Orbital Debris Collision Probability and Maneuver Rate for Space Vehicles,

    James Lee Forster, Jr and Herbert S. Estes, “A Parametric Analysis of Orbital Debris Collision Probability and Maneuver Rate for Space Vehicles,” Aug. 1992

    work page 1992

  8. [8]

    General Method for Calculating Satellite Collision Probability,

    R. P. Patera, “General Method for Calculating Satellite Collision Probability,”Journal of Guidance, Control, and Dynamics, vol. 24, pp. 716–722, July 2001

    work page 2001

  9. [9]

    A Numerical Implementation of Spherical Object Collision Probability,

    S. Alfano, “A Numerical Implementation of Spherical Object Collision Probability,”The Journal of the Astronautical Sciences, vol. 53, pp. 103–109, Mar. 2005

    work page 2005

  10. [10]

    Fast Collision Probability Estimation for Automated Driving using Multi- circular Shape Approximations,

    L. Tolksdorf, C. Birkner, A. Tejada, and N. Van De Wouw, “Fast Collision Probability Estimation for Automated Driving using Multi- circular Shape Approximations,” in2024 IEEE Intelligent Vehicles Symposium (IV), pp. 2529–2536, June 2024. ISSN: 2642-7214

    work page 2024

  11. [11]

    A Particle-Based Collision Probability Estimation Framework for Uncertainty-Aware Risk Evaluation in Autonomous Vehicles,

    A. Hossam, V . Jim´enez-Bermejo, J. Villagra, F. Navas, and V . Milan´es, “A Particle-Based Collision Probability Estimation Framework for Uncertainty-Aware Risk Evaluation in Autonomous Vehicles,”IEEE Access, vol. 14, pp. 39912–39925, 2026

    work page 2026

  12. [12]

    Collision Probability Assessment for Speed Control,

    A. Lambert, D. Gruyer, G. S. Pierre, and A. N. Ndjeng, “Collision Probability Assessment for Speed Control,” in2008 11th International IEEE Conference on Intelligent Transportation Systems, pp. 1043– 1048, Oct. 2008. ISSN: 2153-0017

    work page 2008

  13. [13]

    Probabilistic Collision Checking With Chance Constraints,

    N. E. Du Toit and J. W. Burdick, “Probabilistic Collision Checking With Chance Constraints,”IEEE Transactions on Robotics, vol. 27, pp. 809–815, Aug. 2011

    work page 2011

  14. [14]

    K. V . Mardia and P. E. Jupp, eds.,Directional statistics. Wiley series in probability and statistics, Chichester New York: J. Wiley, 2010

    work page 2010

  15. [15]

    K. V . Mardia, J. T. Kent, and J. M. Bibby,Multivariate analysis. Wiley series in probability and statistics, Hoboken, NJ: Wiley, second edition ed., 2023

    work page 2023

  16. [16]

    CasADi: a software framework for nonlinear optimization and opti- mal control,

    J. A. E. Andersson, J. Gillis, G. Horn, J. B. Rawlings, and M. Diehl, “CasADi: a software framework for nonlinear optimization and opti- mal control,”Mathematical Programming Computation, vol. 11, pp. 1– 36, Mar. 2019

    work page 2019

  17. [17]

    Relations Between Colli- sion Probability, Mahalanobis Distance, and Confidence Intervals for Conjunction Assessment,

    D. Modenini, G. Curzi, and A. Locarini, “Relations Between Colli- sion Probability, Mahalanobis Distance, and Confidence Intervals for Conjunction Assessment,”Journal of Spacecraft and Rockets, vol. 59, pp. 1125–1134, July 2022

    work page 2022