Adaptive fusion localization mechanism towards TDoA and IMU data with LSTM correction Method

Abstract

Using Time Difference of Arrival(TDoA) positioning results and the Inertial measurement unit(IMU) for calculating the motion state of information fusion can significantly improve the positioning accuracy, due to the carrier in the process of movement, the state of the system noise and measurement noise are not strictly obey the normal gaussian distribution, which makes the traditional fusion positioning method using Kalman Filtering algorithm less accurate. This paper proposes an adaptive filter fusion localization mechanism with LSTM network correction. Firstly, a data preprocessing method is designed to convert IMU data from the carrier coordinate system to the geographical coordinate system. Then, based on kinematics theory, the state equation and measurement equation of Adaptive Kalman Filter are established and the system state noise is obtained. Furthermore, the model adaptively to update the carrier coordinate, system state noise and measurement noise. Finally, the carrier trajectory coordinates predicted by the coupled LSTM model are used to obtain the final positioning results and complete the carrier trajectory filtering. Experimental results show that the proposed fusion localization mechanism can effectively improve the accuracy of carrier trajectory localization.