Research On Uncertain Reasoning For AUV Environment Cognition

Since humans had paid more and more attention to the exploitation of marine, the research of Autonomous Underwater Vehicle (AUV) related technology is attracting more and more attention. Among the increasing research of AUV technology, environmental cognition is one of the most important technologies of AUV. Because it guarantees the ability of AUV in performing and the safety and it also decides the level of AUV autonomy. This paper based on the background of AUV, focus on complicated marine environment and AUV status and changeable task and then proposes an environmental cognitive architecture. It studies AUV respectively using information fusion technology Bayesian networks and DS evidence theory in unknown environment for uncertain information how to processing the problem of uncertainty reasoning. The method of the thesis can effectively improve the environmental cognition capability of AUV towards the uncertain environment and improve the autonomy of AUV.Firstly, the research analyzes the background and significance of uncertain reasoning of AUV environment cognition, describes related research in the field of environmental cognitive theory, and elaborates the uncertainties and several commonly used uncertain reasoning methods. Then it gives the main contents of this thesis.Secondly, the AUV architecture, the basis and prerequisite to the future study and design of the AUV functions, gives the related functions and modules dividing as well as the data relationships between each other in overall perspective. From the three kinds of data fusion architecture for information cognitive, combined with the features of AUV about their environment, status and performing task, a task-oriented AUV architecture reference model was given. Based on given reference model according to environment cognitive demands,this paper gives introduce the design and function of state detection module, event detection module and competency assessment module.Thirdly, this paper lays stress on the analysis of AUV uncertain events considering the particularity of AUV environment cognition. Base on the research significance of uncertain events, this paper classified these events according to their reasons. It defined for every uncertain event and focus on different effects from uncertain events, this paper gives different processing solutions.Fourthly, it based on the theoretical background of uncertain reasoning methods with AUV environment cognition, this paper focus on the Bayesian network and DS evidence theory in AUV environment cognition processing application. According to reasoning model of these two methods, each example is given to analyse the reasoning process of the methods.Finally, the results show that the Bayesian network and DS evidence theory can effectively predict these uncertain events through the use of the AUV simulation platform for two uncertain reasoning methods. It improves the ability of environment cognition and autonomy.