Semi-Aerodynamic Model Aided Invariant Kalman Filtering for UAV Full-State Estimation

Xiaoyu Ye,Fujun Song,Zongyu Zhang,Rui Zhang,Qinghua Zeng

2023-10-03

Abstract:Due to the state trajectory-independent features of invariant Kalman filtering (InEKF), it has attracted widespread attention in the research community for its significantly improved state estimation accuracy and convergence under disturbance. In this paper, we formulate the full-source data fusion navigation problem for fixed-wing unmanned aerial vehicle (UAV) within a framework based on error state right-invariant extended Kalman filtering (ES-RIEKF) on Lie groups. We merge measurements from a multi-rate onboard sensor network on UAVs to achieve real-time estimation of pose, air flow angles, and wind speed. Detailed derivations are provided, and the algorithm's convergence and accuracy improvements over established methods like Error State EKF (ES-EKF) and Nonlinear Complementary Filter (NCF) are demonstrated using real-flight data from UAVs. Additionally, we introduce a semi-aerodynamic model fusion framework that relies solely on ground-measurable parameters. We design and train an Long Short Term Memory (LSTM) deep network to achieve drift-free prediction of the UAV's angle of attack (AOA) and side-slip angle (SA) using easily obtainable onboard data like control surface deflections, thereby significantly reducing dependency on GNSS or complicated aerodynamic model parameters. Further, we validate the algorithm's robust advantages under GNSS denied, where flight data shows that the maximum positioning error stays within 30 meters over a 130-second denial period. To the best of our knowledge, this study is the first to apply ES-RIEKF to full-source navigation applications for fixed-wing UAVs, aiming to provide engineering references for designers. Our implementations using MATLAB/Simulink will open source.

Robotics

What problem does this paper attempt to address?

The paper attempts to address the full-state estimation problem of fixed-wing Unmanned Aerial Vehicles (UAVs) under complex flight conditions. Specifically, the study focuses on the following points:

1. **Improving state estimation accuracy and convergence speed**: By introducing the Error-State Right Invariant Extended Kalman Filter (ES-RIEKF) method, it aims to significantly enhance the state estimation accuracy and convergence performance of UAVs when subjected to disturbances.

2. **Multi-sensor data fusion**: A multi-rate sensor network data fusion navigation framework is proposed, combining data from Inertial Measurement Units (IMU), Global Navigation Satellite Systems (GNSS), magnetometers, barometers, airspeed sensors, and control surface deflections to achieve real-time attitude, airflow angle, and wind speed estimation.

3. **Reducing dependence on complex models**: A semi-aerodynamic model fusion framework is introduced, relying only on ground-measurable parameters. A Long Short-Term Memory (LSTM) deep network is designed and trained to achieve drift-free prediction without relying on GNSS or complex aerodynamic model parameters.

4. **Robustness in GNSS-denied environments**: The algorithm's advantages in the absence of GNSS signals are validated. Experimental results show that during a 130-second GNSS-denied period, the maximum positioning error remains within 30 meters.

In summary, the main objective of this paper is to address the full-state estimation problem of fixed-wing UAVs under complex flight conditions by introducing the ES-RIEKF and semi-aerodynamic model-assisted methods, particularly for applications in high-dynamic and GNSS-denied environments.