Moving peak drone search problem: An online multi-swarm intelligence approach for UAV search operations

Many practical, real-world applications have dynamic features. This paper introduces a novel dynamic optimiza-tion problem applied to Unmanned Aerial Vehicle (UAV) search and rescue scenarios, named the Moving Peak Drone Search Problem (MPDSP). It utilizes the dynamic environment generator of the well-known Moving Peak Benchmark (MPB) and it is formulated as a maximization problem, with additional constraints imposed by the use of UAVs. For solving the MPDSP, a multi-swarm framework is proposed and seven optimization algorithms are tested. Five well known swarm intelligence algorithms and two algorithms effectively used in continuous and dynamic optimization problems. The implemented methods are evaluated and compared on 105 scenarios with 4 different UAV fleet configurations. Among the tested swarm intelligence variants, the Particle Swarm Optimization implementations proved to be the most effective for solving the MPDSP.

Automated summary

This paper introduces a novel dynamic optimization problem, Moving Peak Drone Search Problem (MPDSP), and proposes a multi-swarm framework to solve it. By testing seven optimization algorithms, it is found that Particle Swarm Optimization implementations are the most effective for solving the problem.