Research On Navigation And Environment Perception For Unmanned Underwater Vehicle

It is one of the key technique for unmanned underwater vehicle(UUV) to achieve itsautonomous navigation that perceiving the surrounding environment information andlocation itself. The traditional navigation and location algorithm could cause some issuessuch as error accumulation and increased computational complexity. The navigationalgorithm in this paper could reducing the computational complexity, while ensure theaccuracy of navigation, so the study has important theoretical significance and practicalvalue.In this paper, several aspects are studied as follows:Firstly, the dead reckoning algorithm is introduced, the basic principles of deadreckoning is given. And according to the velocity and attitude information measured bysensors that built in the UUV, the dead reckoning of trajectory of UUV is simulatedwithin certain sea area. And the effectiveness of dead reckoning algorithm is verified bycomparing with GPS.Secondly, because of the disadvantage of accumulated error in dead reckoning,extend Kalman filter (EKF) is introduced. The principle of EKF is systematicallydescribed, the key techniques such as feature extraction and map demonstration inenvironment perception. Using a simple UUV motion model, given the coordinate ofinitial position, the trajectory of UUV is predicted with EKF, also the error is analyzed.In addition, the navigation and positioning of UUV based on compressed extendkalman filter (CEKF) is introduced. The disadvantages of it are discussed. The principleand improved method of CEKF are discussed, and main calculation steps are given. Thepartitioning standards and switch criteria of local maps and the data in navigation basedon CEKF is introduced. The CEKF based UUV navigation and positioning is simulatedwith sea trial data. And the effectiveness of CEKF algorithm is verified.Last, in sea trial data verification, compare the results of dead reckoning, EKFmethod, CEKF method and GPS respectively, and the error curve are analyzed. Theresult shows that, CEKF based method has a proper computational complexity, it also hasa higher positioning accuracy than dead reckoning and common EKF based UUVnavigation system.