Spatical And Time Bias Registration For Multisensor Fusion System

Along with the progress of science and technology,single sensor has been unable to meet the requirements of modern warfare in the process of target tracking, multisensor fusion system is needed to improve the performance of target acquisition and tracking. But the bias of the sensors in the fusion system will affect the performance. Therefore, the spatical-time registration is the premise of the fusion system to estimate the state of the moving target correctly. In this paper, the following three aspects are mainly studied.First, in the study of spatical registration of sensors, most previously proposed algorithms need to send the tracks, measurements and the Kalman gains of the local sensor to the fusion center, which leads to the high demand for system data transmission capability. This paper uses an approach that can estimate th e spatical bias without using the local measurements and the Kalman gains. The algorithm firstly uses the inverse Kalman filter to reconstruct the measurements of the local sensors by using the local tracks, then reconstructs the local Kalman gains with the local tracks and measurements, after that the pseudomeasurements of the local sensors is reconstructed, finally designs a filter to estimate the bias vector. This algorithm can reduce the data transmission burden of the fusion system, and effectively estimate the spatical bias vector of the sensor.Secondly, most of the existing time registration algorithms solve the asynchronous problem instead of the inaccurate of the time itself. In this paper, the state augmentation method is generalized, the state ve ctor is extended using the time bias, then the new state vector is sent to the filter to estimate the target state as well as the bias at once. The algorithm estimates bias which is caused by the inaccurate of the time itself, and estimate the the state of the target as well.Finally, most existing algorithms estimate the spatical or time bias assuming the other one was already registrated, but in practical systems, they may exsit at the same time. In this paper, the approach of the second research content was promoted to solve this problem, extending the state vector of the target using the spatical and time bias, then sends the new state vector to the filter to estimate the bias vector. This algorithm estimates the bias and the state of the target at the same time, achieves to estimate the spatical and time bias at once.