Estimation of Distribution Algorithm for Autonomous Underwater Vehicles Path Planning

Abstract

Path planning is that given the start point and target point, finding out a shortest or smallest cost path. In recent years, more and more researchers use evolutionary algorithms (EAs) to solve path planning problems, such as genetic algorithms (GAs) and particle swarm optimization (PSO). Estimation of distribution algorithms (EDAs) belong to a kind of EAs that can make good use of global statistic information of the population. However, EDAs are seldom used to solve path planning problems. In this paper, we propose an EDA variant named adaptive fixed-height histogram (AFHH) algorithm to make path planning for autonomous underwater vehicles (AUVs). The proposed AFHH algorithm can adaptively shrink its search space to make good use of computational resource. We use a regenerate approach to avoid getting stuck in local optimum. We also measure the ability of fixed-height histogram (FHH) algorithm for path planning. We simulate a 3-D environment to measure the ability of the proposed AFHH algorithm. The results show that AFHH has a good convergence rate and can also get better performance.