A multi-modal optimization approach to single path planning for unmanned aerial vehicle

In the past few years, Evolutionary Algorithms (EAs) based UAV path planners have drawn increasing research interests. However, they are not scalable to large-scale problems, i.e., lots of waypoints. Recently, we have proposed a novel EA-based framework, named Separately Evolving Waypoints (SEW), that can deal with large-scale problems. However, the difficulty of UAV path planning depends not only on the number of waypoints, but on the number of constraints it has to satisfy, especially the number of obstacles. In particular, the number of waypoints required is also partly determined by the number of constraints. Hence, it is critical to further improve SEW with respect to large number of obstacles. Originally, a state-of-the-art global optimization approach is employed. In this work, we discuss how the increasing number of obstacles will deteriorate the performance of the global optimizer, then we propose multimodal optimization approaches that facilitates the performance of SEW against large number of obstacles.