Abstract:

For the problem of unmanned aerial vehicle’s multiple routes planning in complex environment, a new method which combines particle swarm optimization (PSO) with weighted k-means clustering is proposed. Each particle represents a route. A weighted k-means clustering algorithm is used to classify the particles to several subgroups. Each subgroup carries out a feasible route optimization. Ultimately multiple different feasible routes are obtained. The traditional k-means clustering algorithm is improved by an exclusion mechanism which generates the initial cluster centers. In order to describe the diversity of unexpected threats distribution in actual environment, route nodes are weighted by the probability of unexpected threat. The weighted k-means clustering algorithm is proposed. Simulation results show that the proposed method can effectively obtain multiple feasible routes.