Researches Into Estimation Of Distribution Algorithm Based-path Planning For Autonomous Underwater Vehicle

With the development of technology,people take more and more attention to the exploring of sea sources.Autonomous underwater vehicle(AUV)is a kind of intelligent robots,which can independently work with the predefined commands.Because of the emergence of AUV,the exploration ability of humans has been further enhanced.Generally,AUV has the ability to plan feasible paths.To enhance its autonomous ability,it is important to research path planning algorithms for AUV.Path planning is to find a path that satisfies some constraints with the given start position and target position in certain environments.The constraints usually include the safety of a path,the smoothness of a path,and the energy consumption.The underwater environment in the sea is very complex,especially the environment close to the seabed.Moreover,the organisms with large body shape and ocean flows may influence the movement of AUV to some degree,and bring challenges of designing path planning algorithms for AUV.Owing to the applications of evolutionary algorithms(EAs)in solving various planning problems in recent years,this paper researches the applications of EA in handling the path planning problem for AUV.As a member of EA,the estimation of distribution algorithm(EDA)is adopted in this paper to plan paths for AUV in both static and dynamic environments.In the static environment,this paper proposes a path planning algorithm based on the EDA for AUV.To improve the accuracy and the speed of finding feasible paths,this paper proposes a strategy that can adaptively shrink the search space.Besides,this paper adopts a periodic disturbance strategy to involve diversity to avoid converging early.The experimental results show that the proposed algorithm has a better performance compared with other EAs.In the dynamic environment,to further develop the path planning algorithm for AUV,this paper proposes a path planning algorithm that can work in both dynamic 2-D and 3-D environments.In this algorithm,a learning transformation strategy is proposed to make individuals learn from the best one in the current population.This strategy can improve the accuracy of the algorithm.Besides,this paper proposes a smooth method to periodically correct individuals in the current population.It is helpful for the algorithm to find feasible paths fast.This paper also adopts a planning window to handle the dynamic factors in the current environment.When the environment changes,the algorithm will find new paths in the planning window.This method can cut down the time consumption to some degree.The experimental results show that the proposed algorithm can effectively handle the path planning problem for AUV in both 2-D and 3-D environments,and it has a better performance compared with other traditional path planning algorithms.