Multisensor Integrated Autonomous Navigation Based on Intelligent Information Fusion

Abstract

A major inadequacy of the Kalman filter is the necessity of accurate measurement models, which may not be possible in the case of complicated disturbances. A new solution to multisensor integrated autonomous navigation based on intelligent information fusion technology for launch vehicles is proposed in this paper, which can improve navigation performance based on the existing measurement models. Artificial neural networks (ANNs) are designed to discover the hidden relationship between the available data and trajectory parameters. The tracking, telemetry, and control system can provide centimeter-level accuracy of the launch vehicle after every flight. While postprocessing data is irrelevant to real-time navigation using traditional methods, it is crucial for ANNs to learn the rules of information fusion and identify the inaccuracy of the input variables. In this way, artificial-intelligence-modified trajectory parameters are expected to be closer to the truth value. The experimental results indicate that the proposed method can significantly improve navigation performance compared with the unscented Kalman filter, and ANNs can achieve good performance in various situations.