Integration of Inertial Navigation System with Global Positioning System using Extended Kalman Filter

The Kalman filter is an optimal state estimator used in many of the engineering applications. In general, Kalman filter has two typical forms for non-linear estimation they are, the extended Kalman filter and unscented Kalman filter. This paper discusses the implementation of GPS-based Inertial navigation system (a combination of accelerometers and gyroscopes) using an extended Kalman filter (EKF). Both GPS and INS sensors are used to update the position and velocity in a land vehicle so that real time guidance and navigation tasks are performed. Due to the low sampling rate of GPS, continuous and reliable position cannot be obtained regularly. Similarly, the INS output tends to accumulate error over time which makes it unreliable. These limitations are overcome by integrating these systems due to their complementary characteristics in order to achieve long term accuracy. The method discussed involves collecting the data from GPS and INS sensors and processing them using EKF to obtain better estimates on the position and velocity. The vehicle orientation is linearized using the quaternion in the prediction and correction stages of EKF.

I. Introduction

Navigation is the process of continuously monitoring and controlling the trajectory of vehicle. Autonomous navigation has garnered a lot of interest in recent years. Land vehicle navigation has been implemented by using various measurement systems like, Global Positioning system (GPS), Inertial Navigation system (INS), Radio navigation system and Vision based navigation. However, the need for accuracy, integrity and robustness has led to integration of these systems to achieve better results. This paper deals with the integration of INS with GPS to predict the states (position, velocity) of a land vehicle. The integration is carried out by an Extended Kalman filter where the readings from the Inertial sensors are used in the Predict stage while the GPS readings are used in the Update stage of the filter.