On-line Path Planning Of AUV Under Dynamic Threat

AUV is an unmanned underwater equipment that is often used as perform military tasks.Therefore,when performing tasks that often require traveling at vital Strait,the main sea channel or the special sea area.In these areas,Dynamic threat exist Frequently.Due to the dynamic threat is agnostic when offline planning,AUV need to avoil the threat dynamically during traveling.Therefore,we need to study the on-line path planning method of AUV under dynamic threat.In terms of handling uncertainty,Bayesian methods make full use of the expert prior knowledge,it does a good work in dealing the combinatorial explosion of the joint probability distribution through its introduction of conditional independence assumptions,Based on the Bayesian approach,this article study how to use Bayesian networks for motion estimation and threat risk assessment and design a online path planning method based on state estimation and risk assessment.Firstly,we introduces the related concepts of Bayesian network,introduces the of Bayesian network inference method and filtering method in detail.And the motion model of underwater moving vehicle is analyzed,deduced the motion model and observation model of dynamic threat.Secondly,according to the motion model of dynamic threat,a double Bayesian motion estimation method is designed.First,calculate the size and direction of the threat's velocity with the observation information,and then use the adaptive Bayesian estimation algorithm to estimate the the size and direction of the dynamic threat's velocity in the next slice of time,and finally use the unscented Bayesian estimation method to estimation the threat's position according to the output of the adaptive Bayesian estimation algorithm before.Compared to the traditional single stage estimation method,the double stage Bayesian estimation method can fit the motion model,and grasp the movement trend of the threat better.Thirdly,because the motion of threat is uncertainty and high mobility,while planning path for AUV,treat the dynamic threat as a static threat at the position whicd the estimation method producted is unreasonable.It should be comprehensive consideration of the threat's movement trend.Aiming at this problem,this paper uses the advantages of Bayesian network inference in dealing with uncertainty problems,design a threat probability evaluation model based on fuzzy discrete dynamic Bayesian network to provide risk assessment on the direction of AUV moves,provides threat information for online path planning for unmanned submersible.Finally,using the ideology “predict the scene-rolling window optimization-feedback and initialization” of model predictive control,make full use of the outpud of the double Bayesian motion estimation and threat probability assessment,design a kind of online path planning method based on the state estimation and the risk assessment.At last,we have simulation tests respectively.in both environment that is no interference and has ocean current in it.