Evaluation of swarm path-planning methods for autonomous area patrolling
Pyykkönen, Pyry (2022)
Pyykkönen, Pyry
2022
Master's Programme in Electrical Engineering
Informaatioteknologian ja viestinnän tiedekunta - Faculty of Information Technology and Communication Sciences
This publication is copyrighted. You may download, display and print it for Your own personal use. Commercial use is prohibited.
Hyväksymispäivämäärä
2022-05-17
Julkaisun pysyvä osoite on
https://urn.fi/URN:NBN:fi:tuni-202204113150
https://urn.fi/URN:NBN:fi:tuni-202204113150
Tiivistelmä
Area patrolling with a swarm of autonomous vehicles is a current research task in the defence and military industries. In order to maximize the area patrolled, the swarm members require decentralized path-planning to ensure both equal division of the patrol area between the swarm members and coverage of the whole area. In this thesis, five decentralized methods for swarm area patrolling is presented. The methods include learning-based methods neuroevolution, deep reinforcement learning and virtual physics optimized with genetical algorithm ,and direct path optimization methods constraint programming and simulated annealing. The thesis evaluates the methods in a custom simulator environment. The direct optimization methods are found to outperform the learning-based methods in both division of the patrol area and providing coverage to a larger area while also providing paths with less obstacles resulting in faster and safer patrolling.