Error Analysis And Compensation Methods Of Leader-follower Cooperative Localization For Multi-USVs

With the development of science and technology, development and utilization of marine resources has become the key point of national attention. Unlike conventional large and expensive surface vessels, unmanned surface vehicles have the features which are small in size, low-cost and autonomously or remote control. So it can execute the missions in unknown or dangerous water area. Unmanned surface vehicle is an offshore platform virtually.With the increasing complexity of marine mission, it is difficult to meet the mission requirements with single platform. People consider to utilize the multiple platforms to collaborate together to complete the mission. The ability of localization is a key part of cooperative operation,so it becomes a popular research area. This paper focuses on unmanned surface vehicle cooperative localization system research.In this thesis, the application of unmanned surface vehicles is the background, Firstly, it briefly introduces the development of unmanned surface vehicles and the research achievements of the cooperative localization method in some country; and then surveys the compensation method of time delays, attitude errors and data dropout in the cooperative localization system. And then, the paper introduces the characteristic of the sensors which are equipped on the Multi-USVs system. Expatiating the principle and the realization process of cooperative localization combine with the equipped sensors. Then using the dead reckoning model of unmanned surface vehicles, this paper establish the system model of the single leader vehicle and multi-leaders vehicles cooperative localization system. Further it gives the algorithm of the cooperative localization based on EKF. And then it carries out the two methods by simulation.According to the USV model, the factors of influencing the cooperative localization accuracy, such as clock uncalibrated and time delay,is analyzed theoretially. Since time delays cause by data transmission and USV movement, residual reconstruction based on state compensation method is deduced. Though reconstructing the residual in time delay situation,then the residual is compensated the estimation to achieve the correction purpose. Then aiming at the problem of large attitude errors which brings out by the low precision micromechanical gyroscope, the paper designs two methods by the quaternion propagation equations. One is the EKF attitude estimation method which utilizes the small-angle approximation. The other is the QKF method based on the model linearized conversion. Due to the communication quality problem, there are measurement data dropouts in cooperative localization system. The standard EKF method has been improved by the random parameter which obeys the Bernoulli distribution to compensate the measurement data dropouts. Also using the same random parameter, the state space model with measurement data dropouts are reconstructed by expanding the state variable. Based on the expanded model, it discusses the minimum variance filter. The single leader vehicle and multi-leaders vehicles cooperative localization methods are validated,and the validation of the methods to compensate the time delay, attitude error and measurement data dropouts are performed by Matlab simulation.