Cooperative Navigation Of Multi-vessels Based On Decentralized Augmented Information Filter

The rapid development of science and technology makes people explore more of themarine field, and a large number of advanced equipments are put into use. Unmanned surfacevessel is becoming an application of hot spots with its lightweight and appropriate price.Because the environment is more complex and subject to many factors, the single vessel withlimited facilities may not be able to complete complex and special task, and only cooperationcan enhance the overall system performance in order to complete the expected task. Based onthe application of decentralized augment information filter, this paper researches the problemof cooperative navigation of multi-vessels.With the application of unmanned surface vessels as the background, the paperintroduces the basic characteristics of the cooperative navigation and the overview of thesensors in detail.Based on the model building, it makes in-depth research of the augmented informationfilter, which is the basis of the decentralized algorithm. Firstly, it describes the parameterforms and sources of the information filters, based on which the paper introduces theaugmented information filter’s concepts and algorithm process. Then based on the Choleskydecomposition it introduces the mean recovery method and several different covariancerecovery methods. Finally, it introduces the concept of Markov random field and informationmatrix, and describes the corresponding relationship of the Markov chain and informationmatrix.Pointing to the augmented information filter’s localized calculated characteristic, thepaper designs its decentralized method of implementation. Firstly, it introduces the local datafusion method and the main content. Then pointing to the dynamic changed joint matrix thepaper introduces a incremental Cholesky revised method, and combining with the distributedcomputing environment it details the distributed Cholesky revised algorithm content. Finally,the above content specific augmented information filer algorithm’s decentralizedimplementation process.The paper introduces decentralized Kalman filtering methods of the cooperativenavigation. Then based on EKF,AIF and DAIF algorithms it designs the simulationexperiments of unmanned surface vessels’ cooperative navigation and real USVs’experimental data of cooperative navigation about DAIF algorithm, and it also verify thecorrectness navigation advantages of the DAIF algorithm by simulation. Finally, in order to verify DAIF algorithm’s superiority in reducing communication and computation, the paperanalyzes accuracy, computation, communication, work-load balance and system stabilities.