A Novel Sensor Fusion Method Based on Invariant Extended Kalman Filter for Unmanned Aerial Vehicle

With the development and popularization of the Artificial Intelligence technology, unmanned aerial vehicles (UAV) are widely used to executing more challenging tasks in complex urban environments, such as delivery, inspection and monitoring. Meanwhile, these complex tasks put forward higher demands for the Navigation system of UAV, and the traditional single sensor such as global positioning system (GPS) and inertial navigation system (INS) have been unable to meet the requirements. In this paper, we derive and implement a novel sensor fusion method based on Invariant Extended Kalman filter (InEKF) with left invariant error (LIEKF) for UAV localization in MATLAB, which inertial measurement unit (IMU) data is used for prediction and GPS data is used for correction. The proposed filter is applied to the simulated data that we generated and also to the Zurich Urban Dataset. By comparing the estimated position to the ground truth, LIEKF has outperformed that of the tradition EKF in both attitude and position in simulated data. It also achieved a better tracking effect and more stable performance in Zurich Urban Dataset, the uniaxial Standard Deviation errors of LIEKF are lower than the traditional EKF.