Unmanned aerial vehicles (UAVs) are becoming an integral part of both industry and society. In particular, the quadrotor is now invaluable across a plethora of fields and recent developments, such as the inclusion of aerial manipulators, only extends their versatility. As UAVs become more widespread, preventing loss-of-control (LOC) is an ever growing concern. Unfortunately, LOC is not clearly defined for quadrotors, or indeed, many other autonomous systems. Moreover, any existing definitions are often incomplete and restrictive. A novel metric, based on actuator capabilities, is introduced to detect LOC in quadrotors. The potential of this metric for LOC detection is demonstrated through both simulated and real quadrotor flight data. It is able to detect LOC induced by actuator faults without explicit knowledge of the occurrence and nature of the failure. The proposed metric is also sensitive enough to detect LOC in more nuanced cases, where the quadrotor remains undamaged but nevertheless losses control through an aggressive yawing manoeuvre. As the metric depends only on system and actuator models, it is sufficiently general to be applied to other systems.

@en

From fault-tolerant control to failure detection, blade damage simulation is integral for developing and testing failure-resilient modern unmanned aerial vehicles. Existing approaches assume partial loss of rotor effectiveness or reduce the problem to centrifugal forces resulting from the shift in the propeller centre of gravity. In this study, a white-box blade damage model based on Blade Element Theory is proposed, integrating both mass and aerodynamic effects of blade damage. The model serves as plug-in to the nominal system model, enables the simulation of any degree of blade damage and does not require costly experimental data from failure cases. A complementary methodology for the identification of the airfoil lift and drag coefficients is also presented. Both contributions were demonstrated with the Bebop 2 drone platform and validated with static test stand wrench measurements obtained at 3 levels of blade damage (0%, 10%, 25%) in a dedicated wind tunnel experimental campaign with velocities up to 12 m/s. Results indicate high accuracy in simulating a healthy propeller. In the presence of blade damage, the model exhibits a relative error between 5% and 24% at high propeller rotational speeds and between 15% and 75% at low propeller rotational speeds.@en

This paper addresses the key question that when faults occur either the aircraft system dynamics changes due to the fault or these dynamics are unknown (precisely). This question is addressed for the important case of Air Data Sensor failures, due to e.g. icing, for fixed wing aircraft operating in a nominal fight condition. The solution to this question uses basic ideas from subspace Identification to cast this problem in linear least squares problem with convex constraints (nuclear norm and 1-norm constraints). The latter are relaxations of a rank and cardinality constraint. The presented solution is validated using real-life fight test data.

@en

Though control algorithms for multirotor Unmanned Air Vehicle (UAV) are well understood, the configuration, parameter estimation, and tuning of flight control algorithms takes quite some time and resources. In previous work, we have shown that it is possible to identify the control effectiveness and motor dynamics of a multirotor fast enough for it to recover to a stable hover after being thrown 4 meters in the air. In this paper, we extend this to include estimation of the position of the Inertial Measurement Unit (IMU) relative to the Center of Gravity (CoG), estimation of the IMU rotation, the thrust direction of all motors and the optimal combined thrust direction. In order to guarantee a correct IMU position estimation, two prior throw-and-catches of the vehicle with spin around different axes are required. For these throws, a height as low as 1 meter is sufficient. Quadrotor flight experimentation confirms the efficacy of the approach, and a simulation shows its applicability to fullyactuated crafts with multiple possible hover orientations. @en

Online fault detection and diagnosis (FDD) enables Unmanned Aerial Vehicles (UAVs) to take informed decisions upon actuator failure during flight, adapting their control strategy or deploying emergency systems. Despite the camera being a ubiquitous sensor on-board of most commercial UAVs, it has not been used within FDD systems before, mainly due to the nonexistence of UAV multi-sensor datasets that include actuator failure scenarios. This paper presents a knowledge-based FDD framework based on a lightweight LSTM network and a single layer neural network classifier that fuses camera and Inertial Measurement Unit (IMU) information. Camera data are pre-processed by first computing its optical flow with RAFT-S, a state-of-the-art deep learning model, and then extracting features with the backbone of MobileNetV3-S. Short-Time Fourier Transform is applied on the IMU data for obtaining their time-frequency information. For training and assessing the proposed framework, UUFOSim was developed: an Unreal Engine-based simulator built on AirSim that allows the collection of high-fidelity photo-realistic camera and sensor information, and the injection of actuator failures during flight. Data were collected in simulation for the Bebop 2 UAV with 16 failure cases. Results demonstrate the added value of the camera and the complementary nature of both sensors with failure detection and diagnosis accuracies of 99.98% and 98.86%, respectively.@en

Ensuring safety in autonomous systems is essential as they become more integrated with modern society. One way to accomplish this is to identify and maintain a safe operating space. To this end, much effort has been devoted in the field of reachability analysis to obtaining control-invariant sets which ensure that a system inside of these sets can remain in these sets, and are thus essential for guaranteeing a system's safety. However, control invariance does not imply that a system can move from any state in the control-invariant set to any other state in the control-invariant set, within a given time horizon. In this paper, we develop an algorithm to obtain a control-invariant set that allows a given system to move from any state in the set to any other state in the set within a given time horizon without having to leave the set. We call this the 'maneuver set', M. We substantiate the algorithm's efficacy through mathematical proof, affirming that the maneuver set obtained through the algorithm is indeed control-invariant. Furthermore, we prove that the system is indeed able to move from any state within this set to any other state in the set. To illustrate the use of our algorithm, we provide the numerical example of a Dubins car, utilising Hamilton-Jacobi-Bellman reachability analysis along with the proposed algorithm in order to obtain M.

@en

This paper presents a method for identifying flight dynamics models for aircraft that includes effects from the flexible structure and the effects from unsteady aerodynamics. In the time domain, the unsteady aerodynamic effects are often modelled using aerodynamic lag states. The proposed method involves first determining the poles that govern the dynamics for these lag states from flight data. This is followed by reconstructing the time signal histories for these lag states so that they can then be used as part of the model fitting procedure. Flight tests were conducted using a scaled Diana2 glider unmanned aerial vehicle (UAV) in order to collect experimental data for modelling. To be able to measure the response of the aircraft and its structure, the glider was instrumented with a wide range of sensors including accelerometers, gyroscopes and strain gauges placed across the aircraft structure. During the flight, various excitation manoeuvres were conducted by the pilot while the aircraft responses were collected. From these measurements, a full flight dynamics model consisting of both lateral and longitudinal dynamics was then identified. Additionally, a model predicting the tail and wing root loads was also identified. First, a rigid aircraft model was fitted that was then extended with states corresponding to the flexible modes and aerodynamic lags. Comparison between the rigid and extended model showed that the addition of structural modes and aerodynamic lag states to the identified models can lead up to 30% improvement in predicting aircraft responses. In conclusion, the method developed and presented in this paper is able to identify flight dynamics models from flight data that more accurately capture the dynamics of flexible aircraft by including effects from the flexible structure and unsteady aerodynamics. @en

This paper proposes an extension to the traditional flight path reconstruction filter to simultaneously reconstruct the aircraft rigid body states together with modal amplitudes and velocities of the structure. To achieve this, the filter makes use of additional accelerometers, gyroscopes and strain gauges placed across the aircraft structure. These measurements are used in a Kalman filter together with kinematic equations of a flexible aircraft and structural mode shapes. First a simulation model is used to evaluate the filter performance on reference signals where the true state is known. Finally, flight test measurements obtained using an instrumented Diana 2 scaled glider are used to validate the filter.@en

Loss of control (LOC) is a prevalent cause of drone crashes. Onboard prevention systems should be designed requiring low computing power, for which data-driven techniques provide a promising solution. This study proposes the use of recurrent neural networks (RNNs) for LOC prediction. Four architectures were trained in order to identify which RNN configuration is most suitable and if this model can predict LOC for changing aerodynamic characteristics, wind conditions, quadcopter types, and LOC events. One-hundred and seventy-two real-world LOC events were conducted using a 53 g Tiny Whoop, a 73 g URUAV UZ85, and a 265 g GEPRC CineGO quadcopter. For these flights, LOC was initiated by demanding an excessive yaw rate (2000 deg/s), which provokes an unrecoverable upset and subsequent crash. All RNNs were trained using only onboard sensor measurements. It was found that the commanded rotor values provided the clearest early warning signals for LOC because these values showed saturation before LOC. Moreover, all four architectures could correctly and reliably predict the impending LOC event 2 s before it actually occurred. Furthermore, to investigate generality of the methodology, the predictors were successfully applied to flight data in which the quadcopter mass, blade diameter, and blade count were varied.@en

Aerodynamic model identification remains essential for simulator operations and control system design and operations. In this paper, state-of-the-art methodologies for aerodynamic model identification and validation are presented, together with a number of novel applications of the identified models that were recently investigated and developed at the Faculty of Aerospace Engineering of Delft University of Technology. In particular, this paper focuses on methodologies for identifying models of aerodynamic stall from flight data, as well as multivariate spline-based aerodynamic model identification methods, together with their applications in flight simulation and advanced flight control.@en

Ensuring the reliability and validity of data-driven quadrotor model predictions is essential for their accepted and practical use. This is especially true for grey- and black-box models wherein the mapping of inputs to predictions is not transparent and subsequent reliability notoriously difficult to ascertain. Nonetheless, such techniques are frequently and successfully used to identify quadrotor models. Prediction intervals (PIs) may be employed to provide insight into the consistency and accuracy of model predictions. This paper estimates such PIs for polynomial and Artificial Neural Network (ANN) quadrotor aerodynamic models. Two existing ANN PI estimation techniques - the bootstrap method and the quality driven method – are validated numerically for quadrotor aerodynamic models using an existing high-fidelity quadrotor simulation. Quadrotor aerodynamic models are then identified on real quadrotor flight data to demonstrate their utility and explore their sensitivity to model interpolation and extrapolation. It is found that the ANN-based PIs widen considerably when extrapolating and remain constant, or shrink, when interpolating. While this behaviour also occurs for the polynomial PIs, it is of lower magnitude. The estimated PIs establish probabilistic bounds within which the quadrotor model outputs will likely lie, subject to modelling and measurement uncertainties that are reflected through the PI widths.@en

In this paper, a new nonlinear control allocation method is presented for a distributed electric propulsion (DEP) aircraft. As the electric propellers can be used actively for control, in addition to the control surfaces, the DEP aircraft is over-actuated. This freedom in control effectors can be exploited with an appropriate control allocation method. All control effectors are, therefore, captured in the incremental nonlinear control allocation (INCA) method, which allows taking into account effector nonlinearities and interactions introduced by the propellers. The INCA method is based on a real-time updated Jacobian model of the control effectiveness, thereby solving an efficient linear control allocation problem. This paper reformulates the original INCA method to optimize the control allocation for minimal propeller power, resulting in more efficient flight. A model predictive control (MPC) controller is added as an actuator dynamics compensation method. This ensures that the commanded control inputs from the INCA controller are achieved. The new controller is compared to a standard incremental nonlinear dynamic inversion (INDI) controller with a translational and rotational loop. It is shown in simulation that by combining INCA with MPC, the tracking performance is improved and efficiency increased by 6.1%.@en

This paper presents a novel method for fast and robust detection of actuator failures on quadrotors. The proposed algorithm has very little model dependency. A Kalman estimator estimates a stochastic effectiveness factor for every actuator, using only onboard RPM, gyro and accelerometer measurements. Then, a hypothesis test identifies the failed actuator. This algorithm is validated online in real-time, also as part of an active fault tolerant control system. Loss of actuator effectiveness is induced by ejecting the propellers from the motors. The robustness of this algorithm is further investigated offline over a range of parameter settings by replaying real flight data containing 26 propeller ejections. The detection delays are found to be in the 30∼130 ms range, without missed detections or false alarms occurring.@en

To gain more insight into human sensitivity to variations in simulated stall buffets, Just Noticeable Difference (JND) thresholds were estimated using a passive human-in-the-loop flight simulator experiment. Using an in-house developed flow separation-based stall and buffet model of the Cessna Citation II, JND thresholds were determined for the model's buffet characteristic frequency parameter omega0 and the buffet onset threshold parameter Xthres for the vertical stall buffet only. With a subjective yes/no 1-up/1-down staircase procedure that uses repeated pairwise comparisons of quasi-steady symmetric stall simulations (where one is a stall with the baseline buffet model and the other one has an offset buffet parameter), upper and lower JND thresholds were measured from 21 pilots. The experiment results show that the pilots noticed the differences in simulated buffet dynamics at comparably similar percentage-wise offsets for Xthres and omega0 with respect to the baseline parameter values. The maximum observed JND thresholds did not exceed 30-35% across all experiment conditions, indicating that pilots are fairly sensitive to even small offsets in the key stall buffet model parameters. Moreover, the estimated JND thresholds for omega0 are in agreement with the +/-2 Hz tolerance currently used in stall buffet simulation qualification standards. However, for Xthres, the results show that human pilots already notice differences in stall buffet onset characteristics well before the maximum allowed tolerance (+/- 2.0 deg angle of attack) is reached, which suggests that stricter tolerances on simulated buffet onsets for quasi-steady symmetric stalls may help to further enhance stall training in simulators. @en

This paper proposes a new method that estimates the three-dimensional stochastic wind velocity for an aircraft equipped with a Pitot-static tube and airflow vanes. Since the performance of most state estimators, e.g., the extended Rauch-Tung-Striebel smoother, relies on the process and measurement noise covariance settings, the proposed method employs the expectation-maximization approach to estimate the noise covariance matrices to improve the estimation accuracy. Numerical simulations demonstrated that the proposed method can successfully estimate the noise covariance matrices, especially for the noise covariance of the wind velocity, using the measurement data and reconstruct the wind velocity offline. Additionally, the smoothed true airspeed, angle of attack, and angle of sideslip data are more accurate compared to the direct measurements. This feature is also beneficial for other applications such as the aerodynamic model identifications of aircraft.@en

Quadcopters are becoming increasingly popular across diverse sectors. Since rotor damages occur frequently, it is essential to improve the attitude estimation and thus ultimately the ability to control a damaged quadcopter. This research is based on a state-of-the-art method that makes it possible to control the quadcopter despite the total failure of a single rotor, where the attitude and position of the quadcopter are provided by an external system. In the present research, a novel attitude estimator called Adaptive Fuzzy Complementary Kalman Filter (AFCKF) has been developed and validated that works independently of any external systems. It is able to estimate the attitude of a quadcopter with one fully damaged rotor while only relying on the on-board MARG (Magnetometer, Accelerometer, Rate Gyroscope) sensors. The AFCKF provides significantly better attitude estimates for flights with a damaged rotor than mainstream filters, estimating the roll and pitch of the quadcopter with an RMS error of less than 1.7 degrees and a variance of less than 2 degrees. The proposed filter also provides accurate yaw estimates despite the fast spinning motion of the damaged quadcopter, and thus outperforms existing methods at the cost of only a small increase in computation.@en

As of 2019, the FAA and EASA require all airline pilots to complete stall and recovery training as integral part of their training. To mitigate risks, this training takes place in ground-based simulators. To enable this, realistic models of aircraft behaviour in the stall regime need to be developed. In this paper, a newaerodynamic stall modelling methodology is proposed that combines classical aerodynamic model identification techniques with a novel adaptation of Kirchhoffs theory of flow separation that considers flow separation over both wings separately. This new model is called the 2X model, as it contains 2 independent flow separation variables, i.e. one for each wing. The model parameters are estimated based on flight experiments in the stall regime conducted with a Cessna Citation II laboratory aircraft operated by the TU-Delft. The developed model for the first time allows accurate prediction of lateral-directional dynamics encountered during stall such as e.g. wing-dip. In addition, it was found that the 2X model also improved predictions of longitudinal stall dynamics leading to a new extended envelope aerodynamic model for the Cessna Citation II that now also includes stall entry and post-stall aerodynamics. @en

Netherlands Aerospace Centre (NLR) and Delft University of Technology (TUD) have obtained an unmanned aerial vehicle to serve as a test platform for future aeroservoelastic flight testing. The purpose of this testbed is to collect data about flexible aircraft flight dynamic responses and loads on the aircraft at unsteady airflow conditions for flexible aircraft system identification. An overview of the 1:3 scaled Diana 2 glider is presented together with numerous tests that were conducted to characterise the test platform. Moment of inertia of the aircraft was determined using a pendulum and rotational swing setup, control surface dynamics were identified from sine sweeps and the structural modes and frequencies were obtained from a ground vibration test. Finally, a low cost and modular data acquisition system was built to collect all the sensor measurements. This data acquisition system is presented together with an overview of its performance.@en

A ground vibration test was conducted with a 1:3 scaled Diana 2 glider model where the modal parameters were estimated using the accelerometers, gyroscopes and strain gauges integrated in the test aircraft and validated using externally attached calibrated accelerometers and commercial software. These modal parameters were then used to update a FEM model of the glider together with static load tests and component mass measurements. The goal for this updated and fitted FEM model is then to build an aeroelastic model for flexible aircraft flight dynamics simulator.

@en

Advancements in aircraft controller design, paired with increasingly flexible aircraft concepts, create the need for the development of novel (smart) adaptive sensing methods suitable for aeroelastic state estimation. A potentially universal and noninvasive approach is visual tracking. However, many tracking methods require manual selection of initial marker locations at the start of a tracking sequence. This study aims to address the gap by investigating a robust machine learning approach for unsupervised automatic labeling of visual markers. The method uses fast DBSCAN and adaptive image segmentation pipeline with hue-saturation-value color filter to extract and label the marker centers under the presence of marker failure. In a comparative study, the DBSCAN clustering performance is assessed against an alternative clustering method, the disjoint-set data structure. The segmentation-clustering pipeline with DBSCAN is capable of running real-time at 250 FPS on a single camera image sequence with a resolution of 1088×600 pixels. To increase robustness against noise, a novel formulation (the inverse DBSCAN, DBSCAN⁻¹ ) is introduced. This approach is validated on an experimental dataset collected from camera observations of a flexible wing undergoing gust excitations in a wind tunnel, demonstrating an excellent match with the ground truth obtained with a laser vibrometer measurement system.

@en