Research On Uncertain Environment Cognition For Unmanned Maritime Systems

Unmanned maritime systems usually execute complex tasks such as navigation,search task and target tracking.In order to enhance the flexibility of unmanned maritime systems under uncertain marine environment,the study of cognitive approaches for uncertain environment has to be carried out.Cognition for uncertain environment is a process to identify uncertain events which may occur in three aspects,i.e.,uncertain marine environment,self-condition of unmanned maritime systems,and mission implementation.Furthermore,the influence degree of these uncertain events on the current tasks of unmanned maritime systems is evaluated.Cognitive results for uncertain environment are used to trigger decision-making module of unmanned maritime systems.The study of cognitive approaches for uncertain environment can effectively offer technical support for the flexibility of unmanned maritime systems,laying its theoretical foundation.The main contents in this paper are as follows:Firstly,the study of cognitive framework of unmanned maritime systems,which includes uncertain information processing,uncertain information regeneration and the knowledge base of uncertain environment,was carried out.The aim is to achieve artificial-intelligent cognitive process for unmanned maritime systems,to uniformly recognize influence degree of various uncertain events on tasks,to dynamically trigger different hierarchic decision-making module by cognitive results from different stages,and to improve the decision-making module efficiency of unmanned maritime systems to the maximum.Compared with other cognitive models of unmanned systems,the developed cognitive model reaches a higher level in three aspects,i.e.,theoretical suitability,engineering applicability and specification.Secondly,the whole system information of cognition for uncertain environment should be represented in the form of knowledge base in an excusable manner by decision-making module.By doing this,the decision-making module of unmanned maritime systems can directly and dynamically obtain cognitive results of uncertain environment of different ranks and types at various stages.Based on the comparative analyses for knowledge expression of other maritime systems,ontology model of unmanned maritime systems for uncertain environment cognition was proposed.The results of simulating experiments indicate that the developed ontology model proves effective and embodies the sharing and reuse of knowledge,interoperability of system.Thirdly,in order to quantify the qualitative results represented by the ontology model at various stages of cognition for uncertain environment and to improve the success rate of cognitive results,the knowledge should be expressed in quantity in the cognitive process according to human thinking,which is accomplished by uncertain reasoning method based on probability.Cognitive approach for uncertain environment of unmanned maritime systems based on the combination of ontology model and Bayesian network was proposed.By implementing concrete tasks of navigation,searching and tracking for AUV under uncertain marine environment,effectiveness of the cognitive approach was validated.Marine and simulating experiments illustrate the effectiveness of the approach.This method could not only solve cognitive problems arising from qualitative and quantitative analyses for uncertain environment,but also make use of the advantages of uncertain reasoning of Bayesian network and domain knowledge sharing of ontology.At last,adjustable autonomous system should be triggered to recalculate autonomous levels,and to assess the operation of users using situation cognitive results.Therefore,situation cognition approach for adjustable autonomous system was proposed in the paper.The system framework of situation cognition approach includes environmental event recognition,influence degree of uncertain events calculation,executive capability assessment,task recommendation and comprehensive situation cognition.The cognition approach for uncertain environment based on the combination of fuzzy logic and Bayesian network was proposed to achieve cognitive process for adjustable autonomous system.The approach was validated by marine experiments and simulating experiments for USV.Results show the effectiveness of situation cognition approach for adjustable autonomous system.