Research of UAV's multiple routes planning based on Multi-Agent Particle Swarm Optimization

To plan multiple routes for unmanned aerial vehicle (UAV), a hybrid algorithm based on multi-agent system (MAS) and particle swarm optimization (PSO) is established, named as Multi-Agent Particle Swarm Optimization (MAPSO). Traditional population structure of the original PSO is adjusted. A particle in MAPSO, being regarded as an agent, represents a candidate route. All agents live in a lattice-like environment, with each agent fixed on a lattice-point. By this means, the speed of information passing among particles is optimized. Moreover, K-means clustering algorithm is introduced to form spatial distinct subpopulations. As a result of all the efforts, an effective way to plan multiple routes is found. Using it, an emulator is designed and some experiments are done. The results prove the feasibility and suitability of the novel method for multiple routes planning issue.