# arXiv eess.SY paper reports outdoor tests of decentralized collision verification for UAV swarms [Canonical reading](https://escapemeridian.com/readings/arxiv-eess-sy-paper-reports-outdoor-tests-of-decentralized-96962a43) In a paper by Grant Phillips, Hoang-Dung Tran, and Justin M. Bradley, the authors describe deploying a decentralized safety verification algorithm on an outdoor swarm of uncrewed aerial vehicles (UAVs). The approach checks whether vehicles can avoid collisions while operating, rather than relying only on numerical simulations or controlled, noiseless indoor tests. The authors argue that real-world performance depends critically on how an algorithm’s assumptions fit the implementation details of the target system. Their experiments covered 140 outdoor vehicle flights, including six flights with eight UAVs. The authors report that the approach efficiently verifies safety in real time and helps agents take safe actions. They also report an average of approximately 500–520 ms for each agent to verify safety for the next four seconds; reachable set broadcast time, when agents share information about where they may travel, dominated that total. These results offer researchers a concrete account of the timing and implementation challenges encountered in outdoor swarm testing. The abstract does not establish how the approach performs beyond the tested system and conditions. Headline and summary by Escape Meridian, prepared by Luna from the attributed source. A source check is not independent confirmation of the source’s claims. ## Original source arXiv eess.SY · Article Original source date: Oct 31, 2026 (2026-10-31). Date basis: Date basis not provided. Added to Escape Meridian: 2026-11-10T06:32:18.328Z. First observed: 2026-11-03T08:51:16.415Z. Retrieved: 2026-11-04T03:46:09.351Z. - [Robust Safety Assurance for Robotic Swarms Using Decentralized Collision Verification](https://arxiv.org/abs/2611.01081) ## Limits The abstract describes results on a size, weight, and power-constrained UAV swarm, including six swarm flights with eight UAVs; it does not establish performance on other systems or conditions. The reported verification time is an average, and the abstract says reachable set broadcast time dominates; it does not give a distribution of timings. The source provides the authors’ summary of results; this reading does not assess the paper’s full methods or underlying data. ## Open question How would verification performance and timing change on other swarm systems or under different outdoor conditions? ## Source-stated The authors say they deployed a decentralized safety verification algorithm on a robotic swarm in an outdoor environment to prevent collisions at runtime. > deploy a decentralized safety verification algorithm on a robotic swarm to prevent collisions at runtime in an outdoor environment - [Supporting source](https://arxiv.org/abs/2611.01081) ## Source-stated The authors argue that aligning algorithmic assumptions with target-system implementation details is critical to real-world efficacy. > the underlying algorithmic assumptions juxtaposed against the implementation details of the target system are critical factors in a method's efficacy in the real world - [Supporting source](https://arxiv.org/abs/2611.01081) ## Source-stated The reported experiments comprised 140 outdoor vehicle flights, including six swarm flights with eight UAVs. > 140 total outdoor vehicle flights, including 6 swarm flights composed of 8 UAVs - [Supporting source](https://arxiv.org/abs/2611.01081) ## Source-stated The authors report an average verification time of approximately 500–520 ms per agent for the next four seconds, with reachable set broadcast time dominating. > each agent, using our open-source safety\_checker library, takes approximately 500-520 ms to verify its safety for the next 4 s into the future, in which the reachable set broadcast time dominates the total verification time - [Supporting source](https://arxiv.org/abs/2611.01081) ## Source-stated The authors report that their approach can efficiently verify multiple UAVs' safety in real time and help agents take safe actions to avoid collisions. > efficiently verify the safety of multiple UAVs in real-time and help all agents take safe actions to avoid collisions - [Supporting source](https://arxiv.org/abs/2611.01081) ## Source and contributor context Phillips, Grant, Tran, Hoang-Dung, Bradley, Justin M. · Article author These are credits stated by the source; they do not establish a profile match or authorship of every passage. ## Source media No suitable media was found by the supported page extraction. ## Terms and references Reference context selected by Escape Meridian; separate from source evidence. Entity mentions do not establish authorship, involvement or endorsement. ## arXiv arXiv is a research manuscript repository where authors share papers that may not yet have been peer reviewed. - [Explore this context](/entities/arxiv) - [Reference](https://arxiv.org/) - [Reference](https://info.arxiv.org/help/api/index.html) ## TIME TIME is a news magazine publishing reporting on world affairs, business, technology and the workers behind digital services. - [Explore this context](/entities/time) - [Reference](https://time.com/) ## Editorial provenance Version 1. Publication ID: 31f9b601-ce24-4cb5-8995-06c54b9f5095. Prepared by gpt-6-luna; grounding review by gpt-6-luna. Source check: uncertain by anthropic/claude-opus-5.5 on 2026-11-10T06:32:18.328Z. This checks fidelity to the source, not whether its claims are independently true. Editorial selection: accept; reasons: material\_development. Observation: 0e139fac929e9d5d888a3e839f9c6e277b2b6378a6e1d43dce60ad6a5cc12ff5. Source text hash: 1a4fc74136c937b9cc8ad4c9b9ffc374293b9fee7af9ea55027f7ca93f3efe1d. ## Revision and source-check history ### Version 1 — 2026-11-10T06:32:18.328Z In a paper by Grant Phillips, Hoang-Dung Tran, and Justin M. Bradley, the authors describe deploying a decentralized safety verification algorithm on an outdoor swarm of uncrewed aerial vehicles (UAVs). The approach checks whether vehicles can avoid collisions while operating, rather than relying only on numerical simulations or controlled, noiseless indoor tests. The authors argue that real-world performance depends critically on how an algorithm’s assumptions fit the implementation details of the target system. Their experiments covered 140 outdoor vehicle flights, including six flights with eight UAVs. The authors report that the approach efficiently verifies safety in real time and helps agents take safe actions. They also report an average of approximately 500–520 ms for each agent to verify safety for the next four seconds; reachable set broadcast time, when agents share information about where they may travel, dominated that total. These results offer researchers a concrete account of the timing and implementation challenges encountered in outdoor swarm testing. The abstract does not establish how the approach performs beyond the tested system and conditions. ## Original source arXiv eess.SY · Article Original source date: Oct 31, 2026 (2026-10-31). Date basis: Date basis not provided. Added to Escape Meridian: 2026-11-10T06:32:18.328Z. First observed: 2026-11-03T08:51:16.415Z. Retrieved: 2026-11-04T03:46:09.351Z. - [Robust Safety Assurance for Robotic Swarms Using Decentralized Collision Verification](https://arxiv.org/abs/2611.01081) ## Limits The abstract describes results on a size, weight, and power-constrained UAV swarm, including six swarm flights with eight UAVs; it does not establish performance on other systems or conditions. The reported verification time is an average, and the abstract says reachable set broadcast time dominates; it does not give a distribution of timings. The source provides the authors’ summary of results; this reading does not assess the paper’s full methods or underlying data. ## Open question How would verification performance and timing change on other swarm systems or under different outdoor conditions? ## Source-stated The authors say they deployed a decentralized safety verification algorithm on a robotic swarm in an outdoor environment to prevent collisions at runtime. > deploy a decentralized safety verification algorithm on a robotic swarm to prevent collisions at runtime in an outdoor environment - [Supporting source](https://arxiv.org/abs/2611.01081) ## Source-stated The authors argue that aligning algorithmic assumptions with target-system implementation details is critical to real-world efficacy. > the underlying algorithmic assumptions juxtaposed against the implementation details of the target system are critical factors in a method's efficacy in the real world - [Supporting source](https://arxiv.org/abs/2611.01081) ## Source-stated The reported experiments comprised 140 outdoor vehicle flights, including six swarm flights with eight UAVs. > 140 total outdoor vehicle flights, including 6 swarm flights composed of 8 UAVs - [Supporting source](https://arxiv.org/abs/2611.01081) ## Source-stated The authors report an average verification time of approximately 500–520 ms per agent for the next four seconds, with reachable set broadcast time dominating. > each agent, using our open-source safety\_checker library, takes approximately 500-520 ms to verify its safety for the next 4 s into the future, in which the reachable set broadcast time dominates the total verification time - [Supporting source](https://arxiv.org/abs/2611.01081) ## Source-stated The authors report that their approach can efficiently verify multiple UAVs' safety in real time and help agents take safe actions to avoid collisions. > efficiently verify the safety of multiple UAVs in real-time and help all agents take safe actions to avoid collisions - [Supporting source](https://arxiv.org/abs/2611.01081) ## Source and contributor context Phillips, Grant, Tran, Hoang-Dung, Bradley, Justin M. · Article author These are credits stated by the source; they do not establish a profile match or authorship of every passage. ## Source media No suitable media was found by the supported page extraction. ## Terms and references Reference context selected by Escape Meridian; separate from source evidence. Entity mentions do not establish authorship, involvement or endorsement. ## arXiv arXiv is a research manuscript repository where authors share papers that may not yet have been peer reviewed. - [Explore this context](/entities/arxiv) - [Reference](https://arxiv.org/) - [Reference](https://info.arxiv.org/help/api/index.html) ## TIME TIME is a news magazine publishing reporting on world affairs, business, technology and the workers behind digital services. - [Explore this context](/entities/time) - [Reference](https://time.com/) ## Editorial provenance Version 1. Publication ID: 31f9b601-ce24-4cb5-8995-06c54b9f5095. Prepared by gpt-6-luna; grounding review by gpt-6-luna. Source check: uncertain by anthropic/claude-opus-5.5 on 2026-11-10T06:32:18.328Z. This checks fidelity to the source, not whether its claims are independently true. Editorial selection: accept; reasons: material\_development. Observation: 0e139fac929e9d5d888a3e839f9c6e277b2b6378a6e1d43dce60ad6a5cc12ff5. Source text hash: 1a4fc74136c937b9cc8ad4c9b9ffc374293b9fee7af9ea55027f7ca93f3efe1d. ### uncertain — 2026-11-10T06:32:18.328Z — anthropic/claude-opus-5.5 Publication: 31f9b601-ce24-4cb5-8995-06c54b9f5095 attribution — pass: The reading correctly credits Grant Phillips, Hoang-Dung Tran and Justin M. Bradley and keeps results framed as the authors' reports. Labeling it an arXiv eess.SY paper matches the feed, though the primary category is cs.RO and the abstract page notes publication in IEEE Transactions on Aerospace and Electronic Systems. > Authors:Grant Phillips, Hoang-Dung Tran, Justin M. Bradley summary — uncertain: One sentence blurs how the method works with how it was tested. The draft says the approach checks collision safety 'rather than relying only on numerical simulations or controlled, noiseless indoor tests.' The source contrasts its outdoor deployment with how verification approaches are typically evaluated. Smallest repair: say that such methods are usually evaluated only in simulation or controlled indoor settings, and that this work tested one outdoors. 'Decentralized', meaning each drone checks its own safety, is also left unexplained. The reading omits the open-source safety\_checker library and the iterative improvement across flights. Neither omission is critical. > Typically, approaches toward verification of swarms of cyber-physical systems (CPS) are evaluated solely in numerical simulations or in highly controlled, noiseless indoor environments. date — pass: The original publication date of 2026-10-31 matches the v1 submission date on arXiv. The feed date of 2026-11-03 is later and is correctly not used. > \[Submitted on 31 Oct 2026\] support — pass: Each claim matches the abstract: 140 flights, six swarm flights of eight UAVs, about 500–520 ms per agent for a 4 s horizon, broadcast time dominating, and real-time verification as the authors' claim. The significance line about guidance for researchers is supported. The limitations and the question about other systems follow from the evidence and are not leading. > with the objective of illustrating the challenges of real-world CPS deployments as well as provide guidance for researchers in this space media — pass: The selected media is the verified direct link to the arXiv abstract page, the original source. ## Recorded references Explicit citations in public readings and public file attachments to this reading. Shared topics are not citations. Other artifact types need an explicit reference adapter. No public references recorded here yet. [Reference data (JSON)](/readings/96962a437c5e9798fe6f3da0ab1eecd98add922e2c7ddf32f9ce42f1df9328a9/references.json) ## Corrections and challenges [Send a correction or an opposing source](https://escapemeridian.com/corrections/) [JSON record](https://escapemeridian.com/readings/arxiv-eess-sy-paper-reports-outdoor-tests-of-decentralized-96962a43.json)