Pith. sign in

REVIEW 3 major objections 7 minor 40 references

Uncertainty Quantification on State-Based Conflict Detection and Resolution Algorithms

T0 review · 3 major / 7 minor · reviewed 2026-08-04 · deepseek-v4-flash

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

Pith's one-line read Under navigation uncertainty, a conflict-resolution algorithm that explicitly maximizes the distance at closest approach is the most robust, and the paper shows Modified Voltage Potential beats Velocity Obstacle in U-space encounters.

desk verdict Good analytical core, useful MVP vs VO comparison, but a factor-2 error in the ADS-L sigma conversion corrupts every absolute quantitative claim. read the letter →

arxiv 2509.10899 v1 pith:MGYJY5KC submitted 2025-09-13 eess.SY cs.SY

Muhammad Fazlur Rahman , Joost Ellerbroek , Jacco Hoekstra This is my paper
classification eess.SYcs.SY
keywords uncertaintyquantificationconflictdetectionandresolutionU-spacenavigationModifiedVoltagePotentialVelocityObstacleclosestpointofapproachintrusionpreventionrate
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

Navigation error turns drone conflict detection and resolution from a deterministic yes/no task into a probabilistic one: the same encounter can be detected or missed depending on sampled position and velocity errors. The paper shows that when the nominal time to intrusion equals the look-ahead horizon, the probability of detecting a conflict falls below 50%. It then compares two resolution algorithms and argues—with analytical approximations, Monte Carlo propagation, and large-scale traffic simulations—that the Modified Voltage Potential (MVP) strategy is more robust because it explicitly maximizes the distance at the closest point of approach, while the Velocity Obstacle (VO) strategy only seeks the smallest velocity change that exits a forbidden cone. The consequence is that at shallow angles and low relative speeds, where VO's admissible region shrinks, MVP maintains higher intrusion-prevention rates and larger post-resolution miss distances. If correct, this establishes a design principle for U-space separation: the optimization criterion, not the size of the allowed velocity change, is what determines safety under uncertainty.

What carries the argument

The central object is dCPA, the vector from the ownship to the predicted closest point of approach (CPA), and its magnitude ||dCPA||, the predicted miss distance. MVP's resolution law aligns the avoidance velocity with dCPA, pushing the aircraft outward from the closest approach. VO instead selects a resolution velocity on the boundary of the velocity obstacle set with minimal deviation from the nominal velocity. The paper also derives the geometry of dCPA under uncertainty: with position noise, dCPA is a Gaussian projection onto the line orthogonal to relative velocity (folded normal magnitude); with velocity noise, dCPA traces a circular arc and its angle follows a projected normal distrib

What would settle it

Re-run the detection and resolution Monte Carlo with sigma_x = sigma_y = 12.25 m and sigma_v = 0.408 m/s (the conventional 95% circular-Gaussian sigmas) and compare detection probabilities and intrusion prevention rates; if MVP's advantage over VO disappears or detection probability at tin = tlookahead no longer falls below 50%, the paper's central quantitative claims are refuted.

Watch

Extended reading notes

Core claim

The paper's central claim is that the most robust conflict resolution strategy under navigation uncertainty is the one that explicitly maximizes ||dCPA||, the predicted distance at the closest point of approach. Under Gaussian position and velocity noise, conflict detection becomes a probabilistic event whose detection probability at the look-ahead boundary is below 50%; repeated observations raise it, but grazing encounters can remain undetected. In resolution, MVP constructs its avoidance vector from dCPA, forcing the aircraft outward from the predicted closest approach, whereas VO picks the minimal velocity deviation to leave the velocity obstacle set. The paper shows that under position

Load-bearing premise

The load-bearing premise is that ADS-L 95% accuracy bounds map to per-axis Gaussian standard deviations of about 6.127 m for a 30 m horizontal bound and 0.204 m/s for a 1 m/s speed bound; under the standard circular-Gaussian interpretation these values are about half the true sigma, so all quantitative probabilities in the paper hinge on this conversion.

Editorial extensions

If this is right

  • When tin equals tlookahead, single-observation conflict detection probability is below 50%, so operational detection must rely on repeated Bernoulli observations; the product of miss probabilities gives the probability of a persistently undetected conflict.
  • MVP's explicit maximization of ||dCPA|| yields intrusion prevention rates close to 1 across heading differences, even under combined position and velocity uncertainty, while VO degrades most at low relative speeds and shallow angles.
  • Look-ahead time and protected-zone radius can be tuned against a target level of safety, e.g., keeping the probability of a 50 m loss of separation below one per million encounters.
  • VO's minimal-change strategy does not improve even when the resolution velocity magnitude is scaled up by 15%, indicating that the optimization criterion, not control authority, is the limiting factor.
  • The paper's conjecture—that the most robust strategy is one that maximizes ||dCPA||—is offered as a foundational design principle for future conflict detection and resolution systems.

Reading between the lines

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

  • The paper's ADS-L conversion appears to understate per-axis sigma by roughly a factor of two (6.127 m and 0.204 m/s versus the usual 95% circular-Gaussian sigma of about 12.25 m and 0.408 m/s); if corrected, the quantitative detection probabilities and IPR values would shift, though the qualitative MVP-over-VO ordering may survive.
  • The CPA-maximization principle could be tested directly against other velocity-space algorithms (such as ORCA or CSORCA) to see whether adding an explicit ||dCPA||-maximizing term to their objective yields the same robustness gains at shallow angles.
  • The arc geometry of dCPA under velocity uncertainty suggests a testable prediction: the probability that VO picks a counterproductive resolution leg should scale with the curvature of that arc, i.e., with inverse relative speed; measuring that correlation in simulation would independently verify the mechanism.
  • The repeated-observation detection model implies a direct relationship between update rate, look-ahead horizon, and target level of safety; a practical extension would be a closed-form formula for the look-ahead needed to achieve P(no detect) below a given threshold as a function of position and velocity noise.
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

3 major / 7 minor

Summary. The paper studies how navigation uncertainty, modelled as zero-mean Gaussian position and velocity errors consistent with ADS-L accuracy bounds, propagates through state-based conflict detection and two decentralized resolution algorithms (Modified Voltage Potential, MVP, and Velocity Obstacle, VO). The authors derive analytical distributions and delta-method approximations for tCPA, ||dCPA||, and tin under position and velocity uncertainty, complement these with Monte Carlo estimates of detection probability, and evaluate macroscopic safety via BlueSky simulations using Intrusion Prevention Rate (IPR) and post-resolution CPA-distance distributions. The central claims are that detection probability falls below 50% when the nominal intrusion time equals the look-ahead, and that MVP, by explicitly maximizing ||dCPA||, is more robust than VO under uncertainty, especially at low relative speeds and shallow angles.

Significance. If the quantitative results are correctly calibrated, the paper makes a useful contribution to U-space safety analysis. It provides closed-form distributional results for detection variables under position uncertainty, carefully documents the breakdown of the delta method under velocity uncertainty, and tests the two algorithms in an independent full-traffic simulator. Strengths include detailed appendices (projection geometry, delta-method gradients, projected-normal angle distribution), explicit labelling of the 'maximize ||dCPA||' principle as a conjecture, and the use of the open-source BlueSky simulator. However, the uncertainty scaling error described below affects all reported numerical magnitudes, so the quantitative claims need to be recomputed and re-examined before the results can be taken as calibrated to the stated ADS-L accuracy.

major comments (3)
  1. [Section 3.2 and Section 4] The conversion of the ADS-L 95% bounds is internally inconsistent by a factor of about two. For a zero-mean circular Gaussian, P(||x||<R)=0.95 implies sigma = R / sqrt(chi2_{2,0.95}) = R/2.4477. With R=30 m this gives sigma ≈ 12.25 m per axis, not 6.127 m; for a 1 m/s 95% speed bound it gives sigma ≈ 0.408 m/s, not 0.204 m/s. The stated values correspond to a 15 m radius and a 0.5 m/s bound. This scaling enters Eqs. (17), (19), (24), (25), (32) and (34), and therefore propagates into the detection probabilities in Fig. 9, the post-resolution dCPA distributions in Figs. 14-19 and 22-24, and the IPR values in Figs. 20-21. Please recompute all quantitative results with the correct scales, or explicitly justify an alternative interpretation of the ADS-L accuracy bounds.
  2. [Sections 5.1.2 and 5.2.2] The paper acknowledges that the delta-method approximations degrade in low-relative-speed, high-uncertainty cases (e.g., Fig. 8, Figs. 18-19, 22-23). These are precisely the regimes where the MVP/VO comparison is most pronounced. The BlueSky Monte Carlo results provide independent evidence, but the analytical support for the proposed mechanism is weakest where the conclusion matters most. The text should state more explicitly which conclusions rest on the MC/BlueSky results rather than on the analytical approximations, and should quantify the MC uncertainty (e.g., standard errors or confidence intervals for the reported detection probabilities and IPRs).
  3. [Section 5.3 / Conclusion] The broader principle, 'the most robust conflict resolution strategy under navigation uncertainty is the one that explicitly maximizes ||dCPA||', is presented as an emerging conclusion in Section 5.3 and the Conclusion, although it is later called a conjecture. Because MVP's resolution vector is constructed directly from dCPA (Eq. 5), the comparison demonstrates the performance of one specific algorithm rather than establishing a general principle. The independent IPR results are consistent with the conjecture but do not prove it. Recommend keeping the conjecture label in the abstract/conclusion wording, not only in the future-work paragraph.
minor comments (7)
  1. [Section 3.1/3.2] The text says the 95% bounds are approximated as ±2 sigma and then uses an inverse-chi-squared circular quantile. These two conventions are not compatible; please make the modelling assumption explicit and consistent.
  2. [Eq. (5)] The MVP resolution formula divides by tCPA · ||dCPA||, which is singular when ||dCPA||=0. Since several experiments use dCPA=0 m, please state the numerical regularization or limiting procedure used in the implementation.
  3. [Section 4.3] Typo: 'This configuration is then repeated simulations are executed' should read 'Each configuration is then repeated in multiple simulations' or similar. Also '50.000 times' should be '50,000 times'.
  4. [Section 5.1.2] Typo: 'This is valid since the the arc-shaped' duplicates 'the'.
  5. [Figures 20-21] IPR values are reported without confidence intervals, despite the text claiming statistical significance from 50,000 repetitions. Please include binomial confidence intervals or at least report the standard error.
  6. [Notation] The symbol dCPA is used both as a vector (Eq. 5) and as a magnitude (Eq. 4, many figure captions). Please use bold or an explicit ||·|| notation consistently to avoid confusion.
  7. [Appendix F] Captions F.25-F.28 quote sigma values (6.127 and 0.204) that depend on the Section 3.2 conversion; these will need to be updated if the conversion is corrected.

Circularity Check

0 steps flagged · score 2.0 of 10

No significant circularity: uncertainty-propagation derivations are self-contained, and the MVP-robustness claim is explicitly labeled a conjecture and tested in independent BlueSky simulations.

full rationale

The paper's analytical derivations (Appendices A–E; delta-method approximations in §5.1) propagate externally specified Gaussian position/velocity uncertainties through the CD&R equations; no fitted parameter is renamed as a prediction. Detection probabilities are empirical Monte Carlo fractions (Eq. 20) computed from those inputs, and the analytical approximations are validated against simulation rather than assumed. The MVP/VO comparison is supported by full BlueSky simulations with IPR (Eq. 21), a metric not part of either algorithm's objective; the claim that maximizing ||dCPA|| is the most robust strategy is explicitly framed as a conjecture in §5.3 and §6 ('rigorously challenge the conjecture that the best conflict resolution algorithm is the one that explicitly maximizes the distance at closest point of approach'), so the paper does not present it as a derived theorem. Self-citations ([5], [15], [18], [19], [29], [36]) supply background, algorithm provenance, and a prior VO variant result; they are not the sole justification for the central uncertainty-propagation results. The MVP algorithm is indeed constructed to push along dCPA (Eq. 5), making the qualitative superiority of MVP on dCPA-based metrics unsurprising, but the paper tests this in simulation and labels the general principle a conjecture, so this is not a case of a prediction being forced by construction. The §3.2 conversion of ADS-L 95% bounds to σx=σy=6.127 m and σv=0.204 m/s appears numerically inconsistent with a 2-D circular-Gaussian 95% radius (which would give approximately 12.25 m and 0.408 m/s), but this is a calibration/correctness concern, not circularity, and does not affect the circularity score.

Assumptions & free parameters 4 free parameters · 5 assumptions · 0 invented entities

The central claims rest on Gaussian zero-mean noise, independence, and the circular-Gaussian interpretation of ADS-L bounds. The delta-method approximations are analytical instruments but fail outside certain regimes, as the paper acknowledges. No new physical entities are introduced.

free parameters (4)
  • position_sigma (per axis) = 6.127 m (stated; likely should be ~12.25 m for a 30 m 95% radius)
    Chosen to represent ADS-L 30 m horizontal position accuracy at 95% confidence; the conversion appears internally inconsistent.
  • velocity_sigma (per axis) = 0.204 m/s (stated; likely should be ~0.408 m/s for a 1 m/s 95% bound)
    Chosen from the 1 m/s ADS-L velocity accuracy bound; the conversion appears internally inconsistent.
  • protected_zone_radius RPZ = 50 m
    Set from operational assumptions in the cited literature, not fitted to data.
  • look_ahead_time tlookahead = 15 s in experiments; 5-7 s in the no-detection example
    Selected from operational conventions; the paper later discusses tuning it to achieve a target level of safety.
assumptions (5)
  • domain assumption Navigation position and velocity errors are zero-mean Gaussian and independent across axes and aircraft.
    Introduced in Section 3.2; this is the core statistical model for all propagation and simulation.
  • domain assumption ADS-L reported accuracy bounds are interpreted as circular Gaussian confidence regions at 95% confidence.
    Stated in Section 3.2; used to derive the per-axis standard deviations.
  • domain assumption State-based conflict detection uses constant-velocity straight-line extrapolation of the current state.
    Defined in Section 2.1; all analytical results are based on this kinematic model.
  • domain assumption Protected zone is a 2D circle and encounters are horizontal.
    Assumed throughout; the paper notes extension to 3D is possible but not treated.
  • standard math First-order delta method provides adequate approximations for the considered non-linear mappings.
    Used for tin under position uncertainty and for tCPA, dCPA, tin under velocity uncertainty; the paper itself notes failure cases.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Uncertainty Quantification on State-Based Conflict Detection and Resolution Algorithms." pith.science (2026). https://pith.science/paper/MGYJY5KC

@misc{pith2026250910899,
  author       = {Pith},
  title        = {Pith review of: Uncertainty Quantification on State-Based Conflict Detection and Resolution Algorithms},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/MGYJY5KC}},
  note         = {Machine review of arXiv:2509.10899}
}
read the original abstract

This study investigates how navigation uncertainty affects conflict detection and resolution (CD&R) for uncrewed aircraft in U-space. Position and velocity errors are modelled as zero-mean Gaussian noise consistent with ADS-L accuracy, and propagated through conflict metrics using Monte Carlo and analytical approximations. Under uncertainty, state-based detection becomes probabilistic. The probability of detection depends on both the level of uncertainty and the encounter geometry, and falls below 50% when the nominal intrusion time equals the look-ahead. Operationally, detection is re-evaluated over time as the encounter develops, yielding multiple observations with varying probabilities. Two resolution algorithms are compared: Modified Voltage Potential (MVP) and Velocity Obstacle (VO). MVP proves more robust under uncertainty because it explicitly maximises distance at the closest point of approach (CPA). By maximising CPA distance, MVP maintains an outward push and avoids reversal behaviour during the manoeuvre, whereas VO performance degrades at low relative speeds and shallow angles. BlueSky simulations confirm these effects: MVP achieves higher intrusion-prevention rates and larger post-resolution miss distances across conflict scenarios, with its advantage most pronounced at low relative velocity. The findings highlight the importance of maximising CPA distance as a conflict resolution strategy. Moreover, the look-ahead horizon and protected zone can be tuned to achieve a desired target level of safety.

Figures

Figures reproduced from arXiv: 2509.10899 by the authors.

Figure 1
Figure 1. State-based conflict detection by calculating the closest point of approach. [PITH_FULL_IMAGE:figures/full_fig_p005_1.png] view at source ↗
Figure 2
Figure 2. Solution space illustration and highlighted Velocity Obstacle (VO) and Modified [PITH_FULL_IMAGE:figures/full_fig_p008_2.png] view at source ↗
Figure 3
Figure 3. Distribution of dCPA under position uncertainty. The spread is constrained to the direction orthogonal to Vrel. σ 2 tin,p ≈ ∇h(µrel) ⊤Σrel ∇h(µrel) (28) The function h(·) is defined in Appendix B as a function of both the time and distance to closest point of approach. The gradient ∇h(µrel) describes how small changes in the relative position affect the estimated time to intru￾sion. The approximation holds only when… view at source ↗
Figures from the paper (21 more)
Figure 4
Figure 4. Figure 4: Probability density comparison between numerical simulation and analytical [PITH_FULL_IMAGE:figures/full_fig_p022_4.png]
Figure 5
Figure 5. Figure 5: Same comparison when ∥dCPA∥ = 45 meters, near the RP Z boundary. Deviation occurs in tin due to samples in dCPA exceeding R2 P Z, where tin becomes undefined. To approximate their mean and variance, we apply the delta method as in the tin calculation for position uncer…
Figure 6
Figure 6. Figure 6: Two-dimensional distribution of dCPA under velocity uncertainty, showing arc￾shaped uncertainty structures for different conflict geometries. 25 [PITH_FULL_IMAGE:figures/full_fig_p025_6.png]
Figure 7
Figure 7. Figure 7: Probability density comparison between numerical simulation and analytical [PITH_FULL_IMAGE:figures/full_fig_p027_7.png]
Figure 8
Figure 8. Figure 8: Same comparison under lower relative velocity: [PITH_FULL_IMAGE:figures/full_fig_p028_8.png]
Figure 9
Figure 9. Figure 9: Detection probability under position and velocity uncertainty [PITH_FULL_IMAGE:figures/full_fig_p030_9.png]
Figure 10
Figure 10. Figure 10: Variation of the time to first intrusion ( [PITH_FULL_IMAGE:figures/full_fig_p031_10.png]
Figure 11
Figure 11. Figure 11: Resolution velocity samples under position uncertainty for [PITH_FULL_IMAGE:figures/full_fig_p035_11.png]
Figure 12
Figure 12. Figure 12: Resolution velocity samples under position uncertainty for [PITH_FULL_IMAGE:figures/full_fig_p036_12.png]
Figure 13
Figure 13. Figure 13: Resolution velocity samples under position uncertainty for [PITH_FULL_IMAGE:figures/full_fig_p037_13.png]
Figure 14
Figure 14. Figure 14: Distribution of post-resolution projected [PITH_FULL_IMAGE:figures/full_fig_p037_14.png]
Figure 15
Figure 15. Figure 15: Distribution of post-resolution projected [PITH_FULL_IMAGE:figures/full_fig_p038_15.png]
Figure 16
Figure 16. Figure 16: Resolution velocity samples under position uncertainty for [PITH_FULL_IMAGE:figures/full_fig_p040_16.png]
Figure 17
Figure 17. Figure 17: Resolution velocity samples under position uncertainty for [PITH_FULL_IMAGE:figures/full_fig_p041_17.png]
Figure 18
Figure 18. Figure 18: Distribution of post- projected |dCP A| for ∆ψ = 2◦ and initial |dCP A| of 0 and 45 meters for velocity uncertainty 41 [PITH_FULL_IMAGE:figures/full_fig_p041_18.png]
Figure 19
Figure 19. Figure 19: Distribution of post-resolution |dCP A| for ∆ψ = 30◦ and initial |dCP A| of 0 and 45 meters for velocity uncertainty all heading differences. When the intruder speed equals the own-ship speed (20 kts), the results diverge: MVP maintains a high IPR for every ∆ψ, wherea…
Figure 20
Figure 20. Figure 20: Intrusion prevention rate under position uncertainty for various intruder speeds. [PITH_FULL_IMAGE:figures/full_fig_p043_20.png]
Figure 21
Figure 21. Figure 21: Intrusion prevention rate under position, velocity and combined uncertainty at [PITH_FULL_IMAGE:figures/full_fig_p044_21.png]
Figure 22
Figure 22. Figure 22: Distribution of ∥dCPA∥ under position uncertainty [PITH_FULL_IMAGE:figures/full_fig_p045_22.png]
Figure 23
Figure 23. Figure 23: Distribution of ∥dCPA∥ under velocity uncertainty both algorithms tend to generate greater separation with more similar dis￾tributions. This occurs because the feasible resolution velocity produced by the VO algorithm, constrained to a circular arc, aligns more closel…
Figure 24
Figure 24. Figure 24: Comparison of ∥dCPA∥ distribution for MVP and VO at different conflict angles under position uncertainty. resolution as the relative velocity increases. Another remark from [PITH_FULL_IMAGE:figures/full_fig_p046_24.png]

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

40 extracted references · 13 canonical work pages

  1. [1]

    Commission, A Drone Strategy 2.0 for a Smart and Sustainable Unmanned Aircraft Eco-System in Europe, Tech

    E. Commission, A Drone Strategy 2.0 for a Smart and Sustainable Unmanned Aircraft Eco-System in Europe, Tech. Rep. COM(2022) 652 final, European Commission, Brussels (Nov. 2022). URLhttps://eur-lex.europa.eu/legal-content/EN/TXT/?uri=celex:52022DC0652

    2022

  2. [2]

    URLhttps://data.europa.eu/doi/10.2829/207917

    SESAR Joint Undertaking, U-space concept of operations (ConOPs): fourth edition., Publications Office, LU, 2023. URLhttps://data.europa.eu/doi/10.2829/207917

    work page doi:10.2829/207917 2023

  3. [3]

    International Civil Aviation Organization (Ed.), Global air traffic man- agement operational concept, 1st Edition, ICAO, Montreal, 2005

    2005

  4. [4]

    European Union Aviation Safety Agency (EASA), Technical Specifica- tion for ADS-L transmissions using SRD-860 frequency band (ADS-L 4 SRD-860), Tech. Rep. ED Decision 2022/024/R, European Aviation Safety Agency, Cologne, Germany (Dec. 2022). URLhttps://www.easa.europa.eu/sites/default/files/dfu/ads-l_4_srd860_issue_1.pdf

    2022

  5. [5]

    M. F. Rahman, J. Ellerbroek, J. M. Hoekstra, Effect of CNS Systems Performance on Autonomous Separation in U-Space, in: E. Neiderman, M. Bourgois, D. Lovell, H. Fricke (Eds.), Proceedings International Conference on Research in Air Transportation, 2024. URLhttps://research.tudelft.nl/en/publications/effect-of-cns-systems-performance-on-autonomous-separati...

    2024

  6. [6]

    Langejan, E

    T. Langejan, E. Sunil, J. Ellerbroek, J. Hoekstra, Effect of ADS-B Characteristics on Airborne Conflict Detection and Resolution, in: 6th SESAR Innovation Days (SID), 2016. URLhttps://research.tudelft.nl/en/publications/effect-of-ads-b-characteristics-on-airborne-conflict-detection-an 59

    2016

  7. [7]

    Z. Khan, H. Idris, M. Consiglio, D. Wing, Surveillance range and inter- ference impacts on self-separation performance, in: 2011 IEEE/AIAA 30th Digital Avionics Systems Conference, 2011, pp. 4D5–1–4D5–16. doi:10.1109/DASC.2011.6095913

    arXiv 2011

  8. [8]

    Weibel, M

    R. Weibel, M. Edwards, C. Fernandes, Establishing a Risk-Based Sepa- ration Standard for Unmanned Aircraft Self Separation, in: 11th AIAA Aviation Technology, Integration, and Operations (ATIO) Conference, American Institute of Aeronautics and Astronautics, Virginia Beach, VA, 2011. doi:10.2514/6.2011-6921. URLhttps://arc.aiaa.org/doi/10.2514/6.2011-6921

    work page doi:10.2514/6.2011-6921 2011

Show all 40 references
  1. [9]

    Ribeiro, J

    M. Ribeiro, J. Ellerbroek, J. Hoekstra, Review of Conflict Resolution Methods for Manned and Unmanned Aviation, Aerospace 7 (6) (2020)

    2020

  2. [10]

    Alejo, R

    D. Alejo, R. Conde, J. Cobano, A. Ollero, Multi-UAV collision avoidance with separation assurance under uncertainties, in: 2009 IEEE International Conference on Mechatronics, 2009, pp. 1–6. doi:10.1109/ICMECH.2009.4957235

    2009

  3. [11]

    Weinert, L

    A. Weinert, L. Alvarez, M. Owen, B. Zintak, Near Midair Collision Analog for Drones Based on Unmitigated Collision Risk, Journal of Air Transportation 30 (2) (2022) 37–48. doi:10.2514/1.D0260. URLhttps://arc.aiaa.org/doi/10.2514/1.D0260

    2022 doi

  4. [12]

    Y. I. Jenie, E.-J. V. Kampen, C. C. De Visser, J. Ellerbroek, J. M. Hoek- stra, Selective Velocity Obstacle Method for Deconflicting Maneuvers Applied to Unmanned Aerial Vehicles, Journal of Guidance, Control, and Dynamics 38 (6) (2015) 1140–1146. doi:10.2514/1.G000737. URLhtt...

    2015 doi

  5. [13]

    L. Yang, J. Kuchar, Using intent information in probabilistic con- flict analysis, in: Guidance, Navigation, and Control Conference and Exhibit, American Institute of Aeronautics and Astronautics, Boston,MA,U.S.A., 1998. doi:10.2514/6.1998-4237. URLhttps://arc.aiaa.org/doi/10....

    1998 doi

  6. [14]

    Granger, N

    G. Granger, N. Durand, A traffic complexity approach through cluster analysis, in: ATM 2003, 5th USA/Europe Air Traffic Management Re- search and Development Seminar, Budapest, Hungary, 2003, p. pp xxxx. URLhttps://enac.hal.science/hal-00938044

    2003

  7. [15]

    Hoekstra, R

    J. Hoekstra, R. Van Gent, R. Ruigrok, Designing for safety: the ‘free flight’ air traffic management concept, Reliability Engineering & System Safety 75 (2) (2002) 215–232. doi:10.1016/S0951-8320(01)00096-5. URLhttps://linkinghub.elsevier.com/retrieve/pii/S0951832001000965

    2002 doi

  8. [16]

    Sunil, J

    E. Sunil, J. Ellerbroek, J. Hoekstra, J. Maas, Modeling Airspace Stabil- ity and Capacity for Decentralized Separation, in: Proceedings of the 12th USA/Europe Air Traffic Management Research and Development Seminar, 2017. URLhttps://repository.tudelft.nl/record/uuid:3e7e2f4a-5...

    2017

  9. [17]

    Fiorini, Z

    P. Fiorini, Z. Shiller, Motion Planning in Dynamic Environments Using Velocity Obstacles, The International Journal of Robotics Research 17 (7) (1998) 760–772. doi:10.1177/027836499801700706. URLhttps://journals.sagepub.com/doi/10.1177/027836499801700706

    1998 doi

  10. [18]

    Balasooriyan, Multi-aircraft Conflict Resolution using Velocity Obstacles, Thesis, Delft University of Technology, Delft (Oct

    S. Balasooriyan, Multi-aircraft Conflict Resolution using Velocity Obstacles, Thesis, Delft University of Technology, Delft (Oct. 2017). URLhttps://repository.tudelft.nl/record/uuid:cf361aff-a7ec-444c-940a-d711e076d108#title

    2017

  11. [19]

    J. M. Hoekstra, J. Ellerbroek, Aerial Robotics: State-based Conflict Detection and Resolution (Detect and Avoid) in High Traffic Densi- ties and Complexities, Current Robotics Reports 2 (3) (2021) 297–307. doi:10.1007/s43154-021-00061-6. URLhttps://link.springer.com/10.1007/s4...

    2021 doi

  12. [20]

    S. B. J. Van Dam, M. Mulder, M. M. van Paassen, Ecological Inter- face Design of a Tactical Airborne Separation Assistance Tool, IEEE Transactions on Systems, Man, and Cybernetics - Part A: Systems and Humans 38 (6) (2008) 1221–1233. doi:10.1109/TSMCA.2008.2001069

    2008

  13. [21]

    Ribeiro, J

    M. Ribeiro, J. Ellerbroek, J. Hoekstra, Velocity Obstacle Based Conflict Avoidance in Urban Environment with Variable Speed Limit, Aerospace 8 (4) (2021) 93. doi:10.3390/aerospace8040093. URLhttps://www.mdpi.com/2226-4310/8/4/93 61

    2021 doi

  14. [22]

    J. P. v. d. Berg, S. J. Guy, M. C. Lin, D. Manocha, Optimal Reciprocal Collision Avoidance for Multi-Agent Navigation, 2010. URLhttps://api.semanticscholar.org/CorpusID:11480907

    2010

  15. [23]

    Durand, Constant speed optimal reciprocal collision avoidance, Transportation Research Part C: Emerging Technologies 96 (2018) 366–379

    N. Durand, Constant speed optimal reciprocal collision avoidance, Transportation Research Part C: Emerging Technologies 96 (2018) 366–379. doi:10.1016/j.trc.2018.10.004. URLhttps://linkinghub.elsevier.com/retrieve/pii/S0968090X18314232

    2018 doi

  16. [24]

    Alligier, D

    R. Alligier, D. Gianazza, N. Durand, X. Olive, Dual-Horizon Reciprocal Collision Avoidance for Aircraft and Unmanned Aerial Systems, Journal of Intelligent & Robotic Systems 107 (1) (2023) 3. doi:10.1007/s10846- 022-01782-2. URLhttps://link.springer.com/10.1007/s10846-022-01782-2

    2023 doi

  17. [25]

    M. S. Eby, A self-organizational approach for resolving air traffic con- flicts, Lincoln Lab. J. 7 (2) (1995) 239–254, place: USA Publisher: Lin- coln Laboratory. URLhttps://dl.acm.org/doi/10.5555/212143.212152

    1995

  18. [26]

    Y. Zou, H. Zhang, G. Zhong, H. Liu, D. Feng, Collision probability estimation for small unmanned aircraft systems, Reliability Engineering & System Safety 213 (2021) 107619. doi:10.1016/j.ress.2021.107619. URLhttps://linkinghub.elsevier.com/retrieve/pii/S0951832021001630

    2021

  19. [27]

    Zhong, S

    G. Zhong, S. Du, H. Zhang, J. Zhou, H. Liu, Demarcation method of safety separations for sUAV based on collision risk estima- tion, Reliability Engineering & System Safety 242 (2024) 109738. doi:10.1016/j.ress.2023.109738. URLhttps://linkinghub.elsevier.com/retrieve/pii/S09518...

    2024

  20. [28]

    Y. I. Jenie, E.-J. Van Kampen, C. C. De Visser, J. Ellerbroek, J. M. Hoekstra, Three-Dimensional Velocity Obstacle Method for Uncoor- dinated Avoidance Maneuvers of Unmanned Aerial Vehicles, Jour- nal of Guidance, Control, and Dynamics 39 (10) (2016) 2312–2323. doi:10.2514/1.G...

    2016 doi

  21. [29]

    M. F. Rahman, A. Vlaskin, J. Ellerbroek, J. Hoekstra, Autonomous Separation in U-Space: Assessing the Impact of Position Uncertainty, 62 SESAR Innovation Days 2024 (2024) 1doi:10.61009/SID.2024.1.42. URLhttps://doi.org/10.61009/SID.2024.1.42

    2024 doi

  22. [30]

    Nister, O

    D. Nister, O. Naroditsky, J. Bergen, Visual odometry, in: Proceed- ings of the 2004 IEEE Computer Society Conference on Computer Vi- sion and Pattern Recognition, 2004. CVPR 2004., Vol. 1, 2004, pp. I–I. doi:10.1109/CVPR.2004.1315094

    2004 arXiv

  23. [31]

    Scaramuzza, F

    D. Scaramuzza, F. Fraundorfer, Visual Odometry [Tutorial], IEEE Robotics & Automation Magazine 18 (4) (2011) 80–92. doi:10.1109/MRA.2011.943233

    2011

  24. [32]

    Delmerico, D

    J. Delmerico, D. Scaramuzza, A Benchmark Comparison of Monocular Visual-Inertial Odometry Algorithms for Flying Robots, in: 2018 IEEE International Conference on Robotics and Automation (ICRA), 2018, pp. 2502–2509. doi:10.1109/ICRA.2018.8460664

    2018

  25. [33]

    Thrun, Probabilistic robotics, Communications of the ACM 45 (3) (2002) 52–57

    S. Thrun, Probabilistic robotics, Communications of the ACM 45 (3) (2002) 52–57. doi:10.1145/504729.504754. URLhttps://dl.acm.org/doi/10.1145/504729.504754

    2002

  26. [34]

    R. A. Paielli, H. Erzberger, Conflict Probability Estimation for Free Flight, Journal of Guidance, Control, and Dynamics 20 (3) (1997) 588–

    1997

  27. [35]

    S. S. Pour, H. Nobahari, M. Prandini, Probability estimation in air- craft conflict detection: a simple and computationally effective method with accuracy certificates, in: 2019 18th European Control Conference (ECC), 2019, pp. 4319–4324. doi:10.23919/ECC.2019.8795909

    2019

  28. [36]

    M. F. Rahman, L. Porricelli, F. Russo, J. Ellerbroek, Valida- tion of e-Conspicuity and Conflict Detection and Resolution Ser- vices in GA-UAS Encounter, in: 2025 Integrated Communica- tions, Navigation and Surveillance Conference (ICNS), 2025, pp. 1–8. doi:10.1109/ICNS65417.2...

    2025

  29. [37]

    Hoekstra, J

    J. Hoekstra, J. Ellerbroek, BlueSky ATC Simulator Project: An Open Data and Open Source Approach, in: 7th International Conference on Research in Air Transportation, 2016. 63 URLhttp://www.icrat.org/icrat/index.cfm, http://www.icrat.org/icrat/7th-international-conference/

    2016

  30. [38]

    K. V. Mardia, P. E. Jupp, Directional statistics, Wiley series in proba- bility and statistics, J. Wiley, Chichester ; New York, 2000. 64

    2000

  31. [79]

    URLhttps://www.mdpi.com/2226-4310/7/6/79

    doi:10.3390/aerospace7060079. URLhttps://www.mdpi.com/2226-4310/7/6/79

    doi

  32. [596]

    URLhttps://arc.aiaa.org/doi/10.2514/2.4081

    doi:10.2514/2.4081. URLhttps://arc.aiaa.org/doi/10.2514/2.4081

    doi

Pith tools

Reviewed August 4, 2026 · model on record in the stance chip above.