Multi-sensor Pose Information Fusion Of A Marine Surface Robot

The marine surface robot that can complete tasks in complex water surface environments has become a global hot research direction.Accurate posture information measurement is an important link to ensure the autonomy of the marine surface robot.Among them,information fusion technology is the key to improve the accuracy and reliability of pose measurement.This thesis uses strapdown inertial navigation system(SINS),global positioning system(GPS)and electronic compass(EC)to establish pose information fusion structure of the marine surface robot,and solves the problem of high-precision pose information fusion of the marine surface robot by optimizing the Kalman filter framework.The main research work is as follows:First,in order to solve the problems of the low accuracy of the pose information of marine surface robot obtained by a single sensor,pose information fusion system of the marine surface robot under loose integration conditions is constructed by proposing an information dynamic sharing federated Kalman filtering.Specifically,starting from the two aspects of the prediction and the update process of the federated Kalman filter,the information sharing principle is improved,further,the D-S evidence theory is used to fuse the error covariance and the measurement noise covariance to optimize the information sharing factor and realize the dynamic adjustment of the information sharing factor.A lot of simulation verification and comparative analysis show that this algorithm can improve the filtering accuracy of the information fusion system,and the root mean square error is reduced by 21.7% compared with the fixed information factor sharing method.Then,in order to solve the problems of unreliable measurement and time-varying noise in the fusion process of the pose information of the marine surface robot under loose integration conditions,a confidence check-adaptive federated Kalman filter framework is proposed.Specifically,by setting the confidence check link,the unreliable measurement information is eliminated and replaced;further,by designing an adaptive adjustment function based on the residual covariance,the adjustment of the system noise covariance matrix is realized.A large number of simulation verifications and comparative analyses show that the introduction of confidence check link can significantly reduce the impact of unreliable measurements on the accuracy of information fusion,at the same time,the adaptive adjustment function can improve the filtering accuracy of the pose information fusion system,and the root mean square error is reduced by 49.5% compared with the Kalman filtering method.Finally,in view of the problem that the accuracy of information fusion under loose integration conditions is limited by the number of GPS satellites,pose information fusion system of the marine surface robot under tight integration conditions is constructed by proposing distributed iterative extended Kalman filter algorithm.Specifically,a distributed structure is used to combine linear and non-linear filters in the fusion system to realize the weight distribution of local filters,by adjusting the Kalman gain and system covariance,the iterative process of the estimated value is constructed to strengthen the convergence performance of the local extended Kalman filter.A large number of simulation verifications and comparative analyses verify the superiority of the proposed algorithm in suppressing the fluctuation of the fusion result,the proposed algorithm improves the accuracy of pose information fusion of the marine surface robot under tight integration conditions,and reduces the root mean square error by 35.1% compared with the extended Kalman filter.In summary,this thesis focuses on the research on the fusion structure of the pose information of the marine surface robot,considers the filtering accuracy and unreliable measurement problems under the loose integration system,a dynamic sharing mechanism for information factors is designed and the confidence check adaptive filtering link is introduced;considers filtering divergence problems under the tight integration system,a weighted distributed structure and iterative strategy are designed.Finally,the filtering accuracy of the pose information fusion system of the marine surface robot is improved.