Deep Kalman Filter: Simultaneous Multi-Sensor Integration and Modelling; A GNSS/IMU Case Study

Abstract

Bayes filters, such as the Kalman and particle filters, have been used in sensor fusion to integrate two sources of information and obtain the best estimate of unknowns. The efficient integration of multiple sensors requires deep knowledge of their error sources. Some sensors, such as Inertial Measurement Unit (IMU), have complicated error sources. Therefore, IMU error modelling and the efficient integration of IMU and Global Navigation Satellite System (GNSS) observations has remained a challenge. In this paper, we developed deep Kalman filter to model and remove IMU errors and, consequently, improve the accuracy of IMU positioning. To achieve this, we added a modelling step to the prediction and update steps of the Kalman filter, so that the IMU error model is learned during integration. The results showed our deep Kalman filter outperformed the conventional Kalman filter and reached a higher level of accuracy.

Keywords

• deep Kalman filter
• Simultaneous Sensor Integration and Modelling (SSIM)
• GNSS/IMU integration
• recurrent neural network (RNN)
• deep learning
• Long-Short Term Memory (LSTM)