Various probabilistic and deterministic methods have been developed and extensively validated in simulation environments to enhance safety in hybrid manned-unmanned airspace. However, reliance on simulation alone limits real-world assessment of algorithm reliability. This work closes a critical simulation-to-flight validation gap for right-of-way (RoW) logic by experimentally evaluating a morphing potential field algorithm in representative encounters and quantifying the minimum detection distance required for a fixed-wing UAS to maintain well-clear from a crewed aircraft. Flight tests were conducted using a general aviation Cessna 172 Skyhawk and a fixed-wing UAS, both instrumented with ADS-B to enable cooperative DAA functionality. Seventeen flight test scenarios were conducted, systematically varying the relative heading and speed to generate a diverse set of encounters and to map well-clear outcomes as a function of detection range. Results indicate that simulations accurately predicted the UAS's ability to maintain the required 2000-ft right-of-way separation in most cases, with 88% of flight test outcomes aligning with simulation-based well-clear classifications. Taken together, the experimentally observed detection range thresholds and the measured simulation-to-flight mismatch demonstrate that high-fidelity nonlinear simulation provides strong predictive capability for DAA performance and can effectively guide design and pre-flight screening. However, the observed sensitivity to nonlinear right-of-way logic, encounter phasing, and unmitigated collision distance highlights the necessity of at least a detection range of 3,962-m or greater in the tested ranges to meet well-clear requirements when utilizing the morphing potential field algorithm.

Real-time planning for fixed-wing unmanned aerial vehicles (UAVs) requires both rapid generation of dynamically feasible trajectories and the ability to reliably track those trajectories. Closed-loop rapidly-exploring random trees (CL-RRT) provide a promising framework by embedding vehicle dynamics and feedback control within rapid tree expansion; however, their application to fixed-wing aircraft remains limited. This work quantifies the impact of aircraft guidance and control architecture on both closed-loop node propagation and reference path tracking in real-time CL-RRT planning. Planning and tracking are first analyzed independently in simulation to isolate architectural effects. Results indicate that controller dynamics shape the short-horizon assessed reachable set during propagation, thereby constraining planner performance and computational robustness. Moreover, fixed-lookahead tracking assumptions break down under the nonuniform spacing and curvature of RRT waypoints. A Total Energy Control System (TECS) scheme and a modified Multi-Segment Adaptive Arc-Length Guidance (MS-AALG) law are proposed to address these shortcomings. Real-time CL-RRT simulations demonstrate improved reliability relative to conventional linear architectures.

In this paper, we present an open-source measurement platform designed to characterize the performance of commercial cellular (Verizon, a major US provider) and LEO satellite (Starlink) networks through real-world flight tests in rural environments. We implement a comprehensive multi-layer measurement approach spanning physical layer signal metrics, multi-cell network topology, and end-to-end (E2E) application performance. Through an extensive flight campaign with more than 10 flight tests, 4.5+ hours of flight time resulting in more than 18K samples, we present the first detailed, open-source dataset analyzing dual cellular and Starlink performance for low-altitude UAV operations. Our cellular-Starlink comparative results, which are collected simultaneously at the same time and location, demonstrate significant performance differences between the two technologies: the LEO satellite link achieves superior latency performance with 95% of Round-Trip Time (RTT) measurements below 50 ms compared to 80% under 150 ms for cellular, and exceptional downlink capacity with 95% exceeding 25 Mbps versus only 5 Mbps for cellular. Our analysis on cellular network performance demonstrates that while higher altitudes (e.g., 330+ m above the sea level) improve signal power by 15-20 dB via line-of-sight (LOS) propagation, it causes a 3-4 × increase in handover rates, which is due to excessive multi-cell visibility rather than signal degradation. Furthermore, we observe asymmetric impacts on the RTT performance due to handovers such that 53.5% of handovers improve RTT, but worst-case degradation (275 ms) is 2 × larger than best-case improvement (137 ms).

Much recent work has focused on the advancement of automatic flight control for autonomous unmanned aerial vehicles, as their low cost and ease of deployment makes them a valuable asset across a wide variety of mission scenarios. This work presents a rate-based optimal control framework for fixed-wing UAS that incorporates actuator rate dynamics directly into the optimization through state and control augmentation. An additional advanced formulation further embeds outer-loop tracking compensation within the optimization to improve steady-state performance under disturbances without requiring cascaded architectures. By modifying the state vector and the control input, the Algebraic Riccati Equation can be solved for the optimal control input to limit the control rates while maintaining high performance. The control designs are validated through real-world flight testing with a wind-to-cruise velocity ratio up to 30%. The rate-based controllers are found to have a 5% improvement in tracking performance compared to standard optimal control schemes while reducing the overall control rates of the aircraft by 16-20%.

Urban air mobility (UAM) operations in dense, obstacle-rich airspace demand collision-avoidance methods that remain safe under tight path constraints and real-world disturbances. Control barrier functions (CBFs) offer a principled framework for safety-critical control, but existing fixed-wing CBF collision-avoidance methods have only been demonstrated in simulation, leaving their guarantees untested under actuator limits, hardware constraints, sensor noise, and atmospheric disturbances. This paper presents the first documented flight-test validation of a higher-order CBF quadratic program (HOCBF-QP) safety filter for fixed-wing UAS collision avoidance. Flight tests are conducted on a SkyHunter aircraft executing a complex figure-8 trajectory in the presence of static obstacles under both low-wind and high-wind conditions. A 3D engagement-dynamics formulation reduces the problem to one-dimensional range dynamics, enabling real-time onboard implementation. A reduced 2D formulation is also derived and compared with the 3D implementation. In low wind, both formulations maintain the required separation throughout all avoidance maneuvers. In high wind, with sustained winds exceeding 40

With the imminent integration of Advanced Air Mobility (AAM) into the national airspace, ensuring the robustness of flight controllers in spatially congested metropolitan areas and in the presence of external disturbances is of paramount importance. The complex interaction between atmospheric turbulence and tall buildings further exacerbates the effects of wind disturbances, posing significant safety challenges to aircraft stability and trajectory tracking. This study employs the high-fidelity urban wind field model using Computational Fluid Dynamics (CFD) which captures wind variations in urban environments. This model quantifies wind shear intensity and vorticity distributions, which are critical factors affecting flight performance. The failure of an autonomous fixed-wing aircraft to maintain its intended flight path within permissible deviation limits under extreme wind conditions is investigated. To address these challenges, a robust flight control system is developed to enhance trajectory tracking performance and mitigate the adverse effects of wind on the path following. The proposed controller is designed to ensure reliable operation despite the unpredictability of urban wind fields, contributing to safer and more resilient autonomous flight operations in complex metropolitan airspaces.

Despite the exponential and promising growth in urban air mobility, this sector faces multifaceted technological and societal challenges. Among the most critical is the development of safe, scalable collision avoidance systems capable of operating within the highly dynamic and congested airspace of metropolitan environments, where complex flight routes must navigate dense infrastructure, variable weather, and unpredictable traffic patterns. Traditional collision avoidance methods, such as potential field methods and TCAS, have limitations at low altitudes and in spatially congested metropolitan areas. This work presents a safety-critical control design using control barrier functions that not only guarantees safe operation but can also be applied to any existing system with minimal impact. Six degrees of freedom simulations show that the controller maintains the ego vehicle's safety across multiple scenarios and is capable of running in real-time for real-world implementation.

A vast body of research exists on the guidance of autonomous unmanned aerial vehicles, with most approaches relying on geometric relationships and constant gains. While these methods can be optimized for predefined flight paths, they become suboptimal in dynamic scenarios requiring real-time guidance without prior knowledge, sharp turns, or significant variations in path length. This work introduces an optimal guidance algorithm with adaptive gains and inherent robustness to external disturbances. By defining the state weighting matrix as a function of cross-track errors, the proposed approach dynamically adjusts gains to minimize deviations. Additionally, incorporating an integral term into the state-space dynamic model ensures zero steady-state error. Lyapunov stability of the algorithm is demonstrated for all possible state weighting matrices. The algorithm is evaluated in a six-degree-of-freedom simulation environment and validated through real-world flight tests under high-wind conditions. Results demonstrate superior robustness and path-tracking performance compared to widely used proportional navigation methods, particularly in adverse environments.

Low-fidelity engineering-level dynamic models are commonly employed while designing uncrewed aircraft flight controllers due to their rapid development and cost-effectiveness. However, during adverse conditions, or complex path-following missions, the uncertainties in low-fidelity models often result in suboptimal controller performance. Aircraft system identification techniques offer alternative methods for finding higher fidelity dynamic models but can be restrictive in flight test requirements and procedures. This challenge is exacerbated when there is no pilot onboard. This work introduces data-driven machine learning (ML) to enhance the fidelity of aircraft dynamic models, overcoming the limitations of conventional system identification. A large dataset from twelve previous flights is utilized within an ML framework to create a long short-term memory (LSTM) model for the aircraft's lateral-directional dynamics. A deep reinforcement learning (RL)-based flight controller is developed using a randomized dynamic domain created using the LSTM and physics-based models to quantify the impact of LSTM dynamic model improvements on controller performance. The RL controller performance is compared to other modern controller techniques in four actual flight tests in the presence of exogenous disturbances and noise, assessing its tracking capabilities and its ability to reject disturbances. The RL controller with a randomized dynamic domain outperforms an RL controller trained using only the engineering-level dynamic model, a linear quadratic regulator controller, and an L1 adaptive controller. Notably, it demonstrated up to 72% improvements in lateral tracking when the aircraft had to follow challenging paths and during intentional adverse onboard conditions.

Urban Air Mobility (UAM) applications, such as air taxis, will rely heavily on perception for situational awareness and safe operation. With recent advances in AUML, state-of-the-art perception systems can provide the high-fidelity information necessary for UAM systems. However, due to size, weight, power, and cost (SWaP-C) constraints, the available computing resources of the on-board computing platform in such UAM systems are limited. Therefore, real-time processing of sophisticated perception algorithms, along with guidance, navigation, and control (GNC) functions in a UAM system, is challenging and requires the careful allocation of computing resources. Furthermore, the optimal allocation of computing resources may change over time depending on the speed of the vehicle, environmental complexities, and other factors. For instance, a fast-moving air vehicle at low altitude would need a low-latency perception system, as a long delay in perception can negatively affect safety. Conversely, a slowly landing air vehicle in a complex urban environment would prefer a highly accurate perception system, even if it takes a little longer. However, most perception and control systems are not designed to support such dynamic reconfigurations necessary to maximize performance and safety. We advocate for developing "anytime" perception and control capabilities that can dynamically reconfigure the capabilities of perception and GNC algorithms at runtime to enable safe and intelligent UAM applications. The anytime approach will efficiently allocate the limited computing resources in ways that maximize mission success and ensure safety. The anytime capability is also valuable in the context of distributed sensing, enabling the efficient sharing of perception information across multiple sensor modalities between the nodes.

In the past three decades, numerous works have been done on the applications of linear and nonlinear Kalman filters in estimating crewed and uncrewed aircraft airflow angles. In uncrewed autonomous aircraft, the flight envelope protection (FEP) algorithms play a vital role in ensuring the safety of the aircraft during flight. The FEP heavily relies on accurate estimations of angle of attack and sideslip angles. In addition to safety, autonomous controller performance can significantly degrade due to poor airflow estimations. Compared to large transport or general aviation aircraft, autonomous aircraft are much lighter, fly lower, and fly slower, which makes them more vulnerable to external disturbances. This work presents the theoretical framework of both linear and nonlinear Kalman filters. It showcases the design process of five different Kalman filters using the $6\text{DoF}$ simulation environment in the presence of sensor noise and external disturbances in the form of the Dryden wind disturbance model. Different Kalman filter designs are assessed using actual flight test data for a realistic evaluation process. Among the five different designs, the Ensemble Kalman filter demonstrated the lowest mean normal of the covariance matrix, indicating superior estimation accuracy. Given the stringent computation power onboard uncrewed aircraft, special attention is given to the computational overhead of each design.

One of the challenges with controlling autonomous aircraft is the coupling between the outer- and inner-loop control blocks, which becomes particularly pronounced when the aircraft executes aggressive maneuvers or experiences adverse onboard conditions such as motor or servo failures. The coupling between the outer- and inner-loop control blocks can cause phase shifts, sustained oscillations, and even lead to loss of control (LoC). To enhance safety and reliability in autonomous aircraft, a unified reinforcement learning (URL) longitudinal control framework has been developed for fixed-wing autonomous aircraft. This strategy replaces the cascaded outer- and inner-loop control blocks with a single inner-outer loop, which calculates the desired control commands (guidance) and executes them (control) in a single step. The structural design of URL enables the integration of fixed-wing aircraft dynamic constraints, such as stall angle of attack and maximum acceleration, as well as physical control constraints, like maximum control surface deflections. These constraints are often challenging to implement in many existing modern control methods. The URL controller robustness to modeling uncertainty is improved using a randomized training environment based on two different LTI models: a physics-based model and a model developed using the derivative-free cross-entropy (CEM) optimization algorithm and flight test data. The validation flight tests demonstrated the URL flight controller's superior performance in real-world environments with low to moderate wind conditions.

Although the autolanding of autonomous aircraft has been widely studied, the majority of research focuses on unconstrained cases where the aircraft has full flexibility to land. Ignoring the physical constraints of the landing area can result in unrealistic navigational paths and even lead to crash landings. This work addresses 4D constraints in autolanding by generating a time-fixed, optimal landing trajectory in three-dimensional space. To guarantee the optimality of solutions, the problem is formulated as an optimal control problem with known initial conditions and a specified trajectory termination point. Simplified aircraft dynamics in the inertial frame are employed to constrain the optimization algorithm, ensuring that only dynamically achievable flight trajectories are generated. In the absence of a priori information about wind heading and landing direction, a sensitivity analysis is conducted on this optimization problem formulation using conditions reflecting real-world flight test data. This analysis reveals that potential energy is the most influential factor for a successful landing trajectory. The robustness of the optimization algorithms is successfully assessed using Monte Carlo analysis, which utilizes Gaussian distributions derived from the flight data, corroborating the findings of the sensitivity analysis.

Since designing aircraft flight controllers is complex, expensive, and time-consuming, the interchangeability of flight controllers between different aircraft platforms has been an active research area. This work presents the development of interchangeable, verifiable flight controllers for fixed-wing UASs. A model-free deep reinforcement learning (RL) algorithm — called PPO (Proximal Policy Optimization) — trains the RL-based control policy. Instead of using high-fidelity dynamic models, the RL policy-based controller is trained in simulation using an engineering-level dynamic model. The robustness of the flight controller toward uncertainty in the dynamic model is improved using randomization of the dynamic model. An aircraft's six degrees of freedom (6DoF) model is used in training to eliminate the heavy reliance of modern controllers on dynamic models, which are prone to the accuracy of the trim information. The idea of an interchangeable flight controller is developed by incorporating memory functions into the policy using long-short-term memory (LSTM), a variant of recurrent neural network (RNN) architecture. The developed flight controller is uniquely verified and validated in actual flight tests using fixed-wing autonomous aircraft. The interchangeable RL-based flight controller is flight-tested on an entirely different aircraft, which is the first of its kind. Its performance is superior to commercial-off-the-shelf flight controllers and LQR-based flight controllers explicitly designed for that platform. Flight test validation and verification data are used to assess flight controller performance and the comparison matrices.

A significant challenge in designing flight controllers lies in their dependency on the quality of dynamic models. This research explores the potential of artificial intelligence-based flight controllers to generalize control actions around policies rather than relying solely on the accuracy of dynamic models. An engineering-level, low-fidelity, linearized model of fixed-wing uncrewed aircraft is used to train a multi-input multi-output (MIMO) flight controller, employing the deep deterministic policy gradients (DDPG) algorithm, to maintain cruise velocity and altitude. While existing literature often concentrates on simulation-based assessments of reinforcement learning (RL)-based flight controllers, this research employs an extensive flight test campaign including 15 flight tests to explore the reliability, robustness, and generalization capability of RL algorithms in tasks they were not specifically trained for, such as changing cruise altitude and velocity. The RL controller outperformed a well-tuned linear quadratic regulator (LQR) on several control tasks. Furthermore, a modification in the DDPG algorithm is presented to enhance the ability of RL controllers to evolve through experience gained from actual flights. The evolved controllers present different behavior compared to the original controller. Comparative flight tests underscored the crucial role of the ratio of actual flight data to the number of simulation-based training instances in optimizing the evolved controllers.

This study explores the application of a comprehensive 3D guidance methodology, developed using control Lyapunov functions, in fixed-wing UAS with significant inertia. Initially, simplified inertia-free particles are utilized for control synthesis, followed by the reintroduction of aircraft dynamics to ensure effective path tracking. In a six-degree-of-freedom simulation environment, the 3D guidance law is compared with the widely adopted $L_{1}$ guidance method, demonstrating exponential convergence to the desired trajectory in both lateral and longitudinal frames. These findings highlight the adaptability and efficacy of control Lyapunov functions, showcasing comparable performance to guidance logics tailored specifically for aerial systems, thus offering promising prospects for precise path following in UAS applications.

This work presents mathematical and practical frameworks for designing deep deterministic policy gradient (DDPG) flight controllers for fixed-wing aircraft. The aim is to design reinforcement learning (RL) flight controllers and accelerate training by substituting the six degrees-of-freedom aircraft models with linear time-invariant (LTI) dynamic models. The initial validation flight tests of the DDPG RL flight controller exhibited poor performance. Post-flight test investigation revealed that the unsatisfactory performance of the RL flight controller could be attributed to the high reliance of the LTI model on accurate control trim values and the substantial errors observed in the predicted trim values generated by the engineering-level dynamic analysis software. A complementary real-time learning Gaussian process (GP) regression was designed to mitigate this critical shortcoming of the LTI-based RL flight controller. The GP estimates and updates the trim control surfaces using observed flight data. The GP regression method incorporates real-time corrections to the trim control surfaces to enhance the performance of the flight controller. Flight test validation was repeated, and the results show that the RL controller, bolstered by the GP trim-finding algorithm, can successfully control the aircraft with excellent tracking performance.

Although unsteady aircraft flight and loss-of-control (LoC) sees extensive study for large transport aircraft, the LoC metrics for uncrewed aerial systems (UAS) are challenging to quantify; however, UAS fly at slower speeds, at lower weights, and small moments of inertia, relatively. This work will provide a method of quantifying when a UAS has entered the nonlinear flight envelope that can lead to LoC by hidden relationships between states using sliding window correlation. Uncrewed aircraft fly significantly closer to ground level (maximum 400 ft AGL) under more influence from turbulent winds. Unlike transport or general aviation (GA) aircraft with a well-developed and restrictive code of federal regulations (CFR), there are no clear regulations for UAS; additionally, UAS have stringent payload capacities limiting the incorporation of redundant systems. The combination of these factors results in UAS having more vulnerability to instability. Aircraft amidst LoC have coupled, unsteady, and nonlinear aerodynamic and propulsive forces and moments; unlike benign flight envelopes, a linearized steady-state model is insufficient. Aircraft in LoC subsist entirely within the nonlinear, unanticipated, and unpredictable flight envelope; therefore, an autopilot system designed around a single linear model will only perpetuate aircraft instability during LoC. To identify the coupled dynamics within a nonlinear flight envelope, the relationships between local aircraft states: angle of attack, sideslip angle, roll rate, pitch rate, and yaw rate, represent the aircraft's stability within the local frame. Control surface deflections provide a control metric against which the aircraft's current dynamic state can gauge. Pattern matching through sliding window correlations is the primary method to find previously negligible hidden relationships during unstable flights. Flight test validation and verification provide ample data for pattern matching; onboard cameras identify when the aircraft has entered instability.

Rapid growth in unmanned aircraft systems (UAS) applications has resulted in exponential increase in the number of new but inexpensive aircraft. Open-source or engineering-level analysis software supports most of these designs and their dynamic analyses. This work analyzes the validity of a perturbed non-linear six-degree-of-freedom simulation of a fixed-wing UAS under six flight conditions. The aircraft model is developed using a component build-up method. Simulations are compared to flight data under different flight conditions: straight flight, level turn, ascending and descending flight. We additionally assessed the dynamic model accuracy when the aircraft was forced into loss of control. In another flight test, the commanded flight speed was reduced to coerce the aircraft into a stall. Unsupervised learning algorithms are used to classify flight data into different flight phases and to select flight portions for analysis. Monte Carlo (MC) simulations are performed to assess dynamic model accuracy while taking simulation parameter uncertainties into account. Results qualify uncertainty levels in predicted states and show that the base dynamic model can only capture aircraft's body rotation rate trends within some errors. The MC simulations mostly capture the flight rotation rates, however, in several instances, the flight data is not captured despite considering simulation parameter uncertainties.