Nonlinear Sequential Fusion Filter And Its Application In Target Trackin

Multi-sensor information fusion techniques are widely used due to the ability of multiple sensors to provide richer information about the observed target.A wellestablished theoretical framework and research methods have been developed for information fusion estimation of linear multi-sensor systems.Among them,sequential fusion estimation methods have received much attention due to their small computational effort and global optimality.However,for non-linear multi-sensor systems,non-linear multi-sensor information fusion is still a hot research topic in the field of information fusion due to the complexity and uncertainty of the non-linear links.In a multi-sensor target tracking system,each sensor cannot receive observations from all sensors simultaneously at each moment due to the difference in transmission time lag or sampling rate,and sequential fusion can process the data sequentially according to the order in which they are received.Based on the Extended Kalman Filter(EKF),this paper investigates the sequential fusion Extended Kalman Filter algorithm and its application in target tracking for a multi-sensor non-linear system,with the following main research elements:A distributed sequential observation fusion estimation algorithm based on the extended Kalman filter is proposed for multi-sensor nonlinear systems.The algorithm reduces the dimensionality of the observation equations of the centralized fusion system and reduces the computational burden of subsequent estimation algorithms.The nonlinear function is linearized by a first-order expansion of the Taylor series,and the observations are sequentially fused in the order of data arrival to obtain the sequential observation fusion estimation algorithm,which is compared with the centralized observation fusion algorithm in terms of estimation accuracy and computational effort.The theoretical and simulation results show that the proposed sequential observation fusion EKF algorithm reduces the computational effort compared to the centralized observation fusion EKF algorithm,and the estimation accuracy is comparable.Based on the extended Kalman filter,matrix full rank decomposition and Weighted Least Squares theory,a sequential weighted observation EKF fusion estimation algorithm is proposed for multi-sensor non-linear systems.When all the sensor observations are compressed,the final compressed observation equations are combined with the state equations to perform extended Kalman filtering,effectively reducing the computational effort of the real-time estimation algorithm.The sequential weighted observation EKF fusion algorithm proposed in this chapter reduces the computational effort compared to the centralized observation fusion algorithm and the estimation accuracy is comparable.The sequential weighted observation EKF fusion algorithm is less computationally intensive than the batch weighted observation fusion algorithm and has the same estimation accuracy.An experimental platform for a two-dimensional target tracking and localization system has been built in the laboratory and the two sequential non-linear filtering algorithms mentioned above have been applied to the experimental platform.Different motion models were developed for robot trajectories in real environments.The base station encapsulates and processes the measured distance information,and then uses the base station module connected to the upper computer to transmit the data to the PC receiver,which processes the data and then inputs the processed data into the MATLAB programming software,thus completing the reception of the data,data fusion processing and display of the positioning tracking effect.The experimental results verify the effectiveness of the proposed sequential fusion estimation algorithm in the application of practical scenarios.