No abstract

A synthetic air data system (SADS) is an analytical redundancy technique that is crucial for unmanned aerial vehicles (UAVs) and is used as a backup system during air data sensor failures. Unfortunately, the existing state-of-theart approaches for SADS require GPS signals or high-fidelity dynamic UAV models. To address this problem, a novel synthetic airspeed estimation method that leverages deep learning and an unscented Kalman filter (UKF) for analytical redundancy is proposed. Our novel fusion-based method only requires an inertial measurement unit (IMU), elevator control input, and airflow angles while GPS, lift/drag coefficients, and complex aircraft dynamic models are not required. Additionally, we demonstrate that our proposed temporal convolutional network (TCN) is a more efficient model for airspeed estimation than the renowned models, such as ResNet or bidirectional long short-term memory (LSTM). Our deep learning-aided UKF was experimentally verified on long-duration real flight data and has promising performance compared with the state-of-theart methods. In particular, it is confirmed that our proposed method robustly estimates the airspeed even under dynamic flight conditions where the performance of conventional methods is degraded.

This paper proposes a novel method for modelaided synthetic airspeed estimation of UAVs. The major contribution of the proposed algorithm is that the synthetic airspeed measurement is newly formulated for analytical redundancy. This filter only requires inertial measurement unit (IMU), airflow angles, and elevator control input along with a simple aircraft model containing only three lift coefficient parameters; no GPS or complex aircraft dynamic model are required. Particularly, two novel filters (unscented Kalman filter and complementary filter) are proposed and evaluated without direct airspeed and GPS measurements. Flight test results of a UAV demonstrated that the proposed algorithm yields accurate estimated airspeed, demonstrating its effectiveness for analytical redundancy.

Identifying the drag parameters of UAVs is crucial for guaranteeing their aerodynamic efficiency. However, in contrast to commercial aircraft, obtaining accurate UAV drag parameters is challenging since the existing approaches rely on analytical models or require accurate modeling of the engine thrust, which highly depends on time-varying wind conditions. To address this challenge, this paper first proposes a novel in-flight estimation algorithm for the air data (airspeed, angle of attack, and sideslip angle) and drag parameters of UAVs. With this approach, there is no need to compute all of the contributing components for drag, to model the thrust of the UAVs or to perform complicated wind tunnel testing/computational fluid dynamics (CFD) analysis to obtain the drag parameter. Instead, the proposed algorithm requires only standard sensors such as inertial measurement units (IMUs) and air data systems during gliding flight. Then, an efficient glide phase detection algorithm for initiating the filter is proposed. Simulation and experimental flight testing of a UAV demonstrate that the proposed algorithm yields accurate zero lift drag coefficient, attitude, and air data estimation results according to thorough validation with newly derived metrics for performance evaluation.

This paper presents an algorithm for autonomous landing approach of a unmanned aerial vehicle. The main purpose of the autonomous landing approach in this study is to help a safe landing at night. From any initial position of the aircraft when this function is engaged, a flight path command is generated from the initial position. The shortest combination of an initial circular arc, a straight line segment, and a final circular arc is chosen for the flight path that will lead the aircraft to one end of runway for a landing. The algorithm is initially validated through numerous simulations with various initial conditions of aircraft. Then it is successfully validated through a number of flight tests.