Neural Tuning of Parameters in a Collision Avoidance Mechanism for Swarms of Drones

Dariusz Marek; Jakub Szyguła; Piotr Biernacki; Marcin Paszkuta; Adam Domański; Marcin Fojcik; Marta Szczygieł; Konrad Wojciechowski

doi:10.1145/3737899.3768522

Neural Tuning of Parameters in a Collision Avoidance Mechanism for Swarms of Drones

Dariusz Marek
, Jakub Szyguła
, Piotr Biernacki
, Marcin Paszkuta
, Adam Domański
, Marcin Fojcik
, Marta Szczygieł
, Konrad Wojciechowski

Institute of Theoretical and Applied Informatics of the Polish Academy of Sciences
Western Norway University of Applied Sciences
Polish-Japanese Academy of Information Technology
Academy of Silesia

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › peer-review

1 Citation (Scopus)

Abstract

This paper presents a simplified mechanism for collision avoidance in drone swarms flying in dynamic environments. The algorithm is based on sharing information between drones about their positions and planned movements, and uses repulsion vectors to adjust flight paths and avoid obstacles. In this work, we introduce a way to automatically adjust one of the key parameters of the algorithm using a simple artificial neuron. This makes the system more flexible and better adapted to changing conditions, while still keeping it easy to implement and suitable for use in small, low-cost drones. The solution works without central control and supports safe, coordinated flight.

Original language	English
Title of host publication	FLEdge-AI 2025 - Proceedings of the 2025 Federated Learning and Edge AI for Privacy and Mobility
Publisher	Association for Computing Machinery, Inc
Pages	44-51
Number of pages	8
ISBN (Electronic)	9798400719769
DOIs	https://doi.org/10.1145/3737899.3768522
Publication status	Published - 2 Dec 2025
Event	2025 Federated Learning and Edge AI for Privacy and Mobility, FLEdge-AI 2025 - Hong Kong, China Duration: 4 Nov 2025 → 8 Nov 2025

Publication series

Name	FLEdge-AI 2025 - Proceedings of the 2025 Federated Learning and Edge AI for Privacy and Mobility

Conference

Conference	2025 Federated Learning and Edge AI for Privacy and Mobility, FLEdge-AI 2025
Country/Territory	China
City	Hong Kong
Period	4/11/25 → 8/11/25

UN SDGs

This output contributes to the following UN Sustainable Development Goals (SDGs)

SDG 3 Good Health and Well-being

Keywords

collision avoidance
neural networks, distributed systems, Federated Learnings, UAVs, positioning accuracy
software-in-the-loop (SITL)
Swarm of Drones

ASJC Scopus subject areas

Artificial Intelligence
Computer Science Applications
Electrical and Electronic Engineering
Anesthesiology and Pain Medicine

Access to Document

10.1145/3737899.3768522

Cite this

Marek, D., Szyguła, J., Biernacki, P., Paszkuta, M., Domański, A., Fojcik, M., Szczygieł, M., & Wojciechowski, K. (2025). Neural Tuning of Parameters in a Collision Avoidance Mechanism for Swarms of Drones. In FLEdge-AI 2025 - Proceedings of the 2025 Federated Learning and Edge AI for Privacy and Mobility (pp. 44-51). (FLEdge-AI 2025 - Proceedings of the 2025 Federated Learning and Edge AI for Privacy and Mobility). Association for Computing Machinery, Inc. https://doi.org/10.1145/3737899.3768522

Marek, Dariusz ; Szyguła, Jakub ; Biernacki, Piotr et al. / Neural Tuning of Parameters in a Collision Avoidance Mechanism for Swarms of Drones. FLEdge-AI 2025 - Proceedings of the 2025 Federated Learning and Edge AI for Privacy and Mobility. Association for Computing Machinery, Inc, 2025. pp. 44-51 (FLEdge-AI 2025 - Proceedings of the 2025 Federated Learning and Edge AI for Privacy and Mobility).

@inproceedings{a630d86e208b42da8a6fd4b9a1a6676c,

title = "Neural Tuning of Parameters in a Collision Avoidance Mechanism for Swarms of Drones",

abstract = "This paper presents a simplified mechanism for collision avoidance in drone swarms flying in dynamic environments. The algorithm is based on sharing information between drones about their positions and planned movements, and uses repulsion vectors to adjust flight paths and avoid obstacles. In this work, we introduce a way to automatically adjust one of the key parameters of the algorithm using a simple artificial neuron. This makes the system more flexible and better adapted to changing conditions, while still keeping it easy to implement and suitable for use in small, low-cost drones. The solution works without central control and supports safe, coordinated flight.",

keywords = "collision avoidance, neural networks, distributed systems, Federated Learnings, UAVs, positioning accuracy, software-in-the-loop (SITL), Swarm of Drones",

author = "Dariusz Marek and Jakub Szygu{\l}a and Piotr Biernacki and Marcin Paszkuta and Adam Doma{\'n}ski and Marcin Fojcik and Marta Szczygie{\l} and Konrad Wojciechowski",

note = "Publisher Copyright: {\textcopyright} 2025 Copyright held by the owner/author(s).; 2025 Federated Learning and Edge AI for Privacy and Mobility, FLEdge-AI 2025 ; Conference date: 04-11-2025 Through 08-11-2025",

year = "2025",

month = dec,

day = "2",

doi = "10.1145/3737899.3768522",

language = "English",

series = "FLEdge-AI 2025 - Proceedings of the 2025 Federated Learning and Edge AI for Privacy and Mobility",

publisher = "Association for Computing Machinery, Inc",

pages = "44--51",

booktitle = "FLEdge-AI 2025 - Proceedings of the 2025 Federated Learning and Edge AI for Privacy and Mobility",

}

Marek, D , Szyguła, J , Biernacki, P , Paszkuta, M , Domański, A, Fojcik, M, Szczygieł, M & Wojciechowski, K 2025, Neural Tuning of Parameters in a Collision Avoidance Mechanism for Swarms of Drones. in FLEdge-AI 2025 - Proceedings of the 2025 Federated Learning and Edge AI for Privacy and Mobility. FLEdge-AI 2025 - Proceedings of the 2025 Federated Learning and Edge AI for Privacy and Mobility, Association for Computing Machinery, Inc, pp. 44-51, 2025 Federated Learning and Edge AI for Privacy and Mobility, FLEdge-AI 2025, Hong Kong, China, 4/11/25. https://doi.org/10.1145/3737899.3768522

Neural Tuning of Parameters in a Collision Avoidance Mechanism for Swarms of Drones. / Marek, Dariusz ; Szyguła, Jakub ; Biernacki, Piotr et al.
FLEdge-AI 2025 - Proceedings of the 2025 Federated Learning and Edge AI for Privacy and Mobility. Association for Computing Machinery, Inc, 2025. p. 44-51 (FLEdge-AI 2025 - Proceedings of the 2025 Federated Learning and Edge AI for Privacy and Mobility).

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › peer-review

TY - GEN

T1 - Neural Tuning of Parameters in a Collision Avoidance Mechanism for Swarms of Drones

AU - Marek, Dariusz

AU - Szyguła, Jakub

AU - Biernacki, Piotr

AU - Paszkuta, Marcin

AU - Domański, Adam

AU - Fojcik, Marcin

AU - Szczygieł, Marta

AU - Wojciechowski, Konrad

PY - 2025/12/2

Y1 - 2025/12/2

N2 - This paper presents a simplified mechanism for collision avoidance in drone swarms flying in dynamic environments. The algorithm is based on sharing information between drones about their positions and planned movements, and uses repulsion vectors to adjust flight paths and avoid obstacles. In this work, we introduce a way to automatically adjust one of the key parameters of the algorithm using a simple artificial neuron. This makes the system more flexible and better adapted to changing conditions, while still keeping it easy to implement and suitable for use in small, low-cost drones. The solution works without central control and supports safe, coordinated flight.

AB - This paper presents a simplified mechanism for collision avoidance in drone swarms flying in dynamic environments. The algorithm is based on sharing information between drones about their positions and planned movements, and uses repulsion vectors to adjust flight paths and avoid obstacles. In this work, we introduce a way to automatically adjust one of the key parameters of the algorithm using a simple artificial neuron. This makes the system more flexible and better adapted to changing conditions, while still keeping it easy to implement and suitable for use in small, low-cost drones. The solution works without central control and supports safe, coordinated flight.

KW - collision avoidance

KW - neural networks, distributed systems, Federated Learnings, UAVs, positioning accuracy

KW - software-in-the-loop (SITL)

KW - Swarm of Drones

UR - https://www.scopus.com/pages/publications/105025598783

U2 - 10.1145/3737899.3768522

DO - 10.1145/3737899.3768522

M3 - Conference contribution

AN - SCOPUS:105025598783

T3 - FLEdge-AI 2025 - Proceedings of the 2025 Federated Learning and Edge AI for Privacy and Mobility

SP - 44

EP - 51

BT - FLEdge-AI 2025 - Proceedings of the 2025 Federated Learning and Edge AI for Privacy and Mobility

PB - Association for Computing Machinery, Inc

T2 - 2025 Federated Learning and Edge AI for Privacy and Mobility, FLEdge-AI 2025

Y2 - 4 November 2025 through 8 November 2025

ER -

Marek D , Szyguła J , Biernacki P , Paszkuta M , Domański A, Fojcik M et al. Neural Tuning of Parameters in a Collision Avoidance Mechanism for Swarms of Drones. In FLEdge-AI 2025 - Proceedings of the 2025 Federated Learning and Edge AI for Privacy and Mobility. Association for Computing Machinery, Inc. 2025. p. 44-51. (FLEdge-AI 2025 - Proceedings of the 2025 Federated Learning and Edge AI for Privacy and Mobility). doi: 10.1145/3737899.3768522

Neural Tuning of Parameters in a Collision Avoidance Mechanism for Swarms of Drones

Abstract

Publication series

Conference

UN SDGs

Keywords

ASJC Scopus subject areas

Access to Document

Other files and links

Fingerprint

Cite this