Features of Invariant Extended Kalman Filter Applied to Unmanned Aerial Vehicle Navigation

Area of Science:

• Robotics and Control Systems
• Aerospace Engineering
• Navigation Systems

Background:

• Kalman filters are essential for state estimation in autonomous systems.
• Extended Kalman Filters (EKF) are widely used but can suffer from divergence.
• Invariant Extended Kalman Filters (IEKF) present a newer, potentially more robust alternative.

Purpose of the Study:

• To evaluate the practical performance of an Invariant Extended Kalman Filter (IEKF) for Unmanned Aerial Vehicle (UAV) navigation.
• To compare the IEKF's stability and convergence against a traditional Extended Kalman Filter (EKF) based navigation method.
• To analyze the distinct features and internal parameters of both IEKF and EKF in a real-world flight test scenario.

Main Methods:

• Implementation of an IEKF algorithm for UAV navigation.
• Comparison with an open-source EKF-based navigation system.
• Utilized sensor data from GPS, MEMS-AHRS (Microelectromechanical system-attitude heading reference system), and barometric altimeter.
• Conducted flight tests using both rotary-wing and fixed-wing UAVs.

Main Results:

• IEKF demonstrated superior stability and convergence in estimated states and internal parameters (Kalman gain, state error covariance, measurement innovation).
• Location, velocity, and altitude estimations were comparable between IEKF and EKF methods.
• Internal parameter analysis revealed key differences in the filtering behavior.

Conclusions:

• The IEKF provides a more robust and stable navigation solution for UAVs compared to the conventional EKF.
• IEKF's enhanced stability is evident in its internal filtering dynamics, even when external state estimations are similar.
• This research validates the practical benefits of IEKF for real-world UAV applications.