Keywords: Aerial Robotic Manipulation, Control Architectures, Multirotor Design and Control
Abstract: Identifying predictive models of nonlinear dynamical systems directly from visual observations remains a fundamental challenge, particularly in the context of control. In this paper, we propose a deep learning–based Koopman identification framework that learns control-oriented models of nonlinear dynamical systems from visual data. In contrast to existing Koopman-based approaches, our method incorporates additional geometric consistency losses and represents the lifted system dynamics using both linear and bilinear model formulations. Closed-loop trajectory-tracking simulations of a quadrotor observed by an external camera demonstrate the model's capacity to accurately capture the underlying system dynamics, enabling reliable visual predictions and effective visual control.

Keywords: Autonomy, Control Architectures, Path Planning
Abstract: This paper presents a hierarchical decision-making framework for unmanned aerial vehicle (UAV) missions motivated by search-and-rescue (SAR) scenarios under limited simulation training. The framework combines a fixed rule-based high-level advisor with an online goal-conditioned low-level reinforcement learning (RL) controller. To stress-test early adaptation, we also consider a strict no-pretraining deployment regime. The high-level advisor is defined offline from a structured task specification and compiled into deterministic rules. It provides interpretable mission- and safety-aware guidance through recommended actions, avoided actions, and regime-dependent arbitration weights. The low-level controller learns online from task-defined dense rewards and reuses experience through a mode-aware prioritized replay mechanism augmented with rule-derived metadata. We evaluate the framework on two tasks: battery-aware multi-goal delivery and moving-target delivery in obstacle-rich environments. Across both tasks, the proposed method improves early safety and sample efficiency primarily by reducing collision terminations, while preserving the ability to adapt online to scenario-specific dynamics.

Keywords: Control Architectures, Simulation, Autonomy
Abstract: This paper presents a deep Q-network (DQN)–based gain-scheduling framework for safety-critical quadcopter trajectory tracking. Instead of directly learning control inputs, the proposed approach selects from a finite set of pre-certified stabilizing gain vectors, enabling reinforcement learning to operate within a structured and stability-preserving control architecture. By exploiting the isotropic structure of the translational dynamics, feedback gains are shared across spatial axes to reduce dimensionality while preserving performance. The learned policy adapts feedback aggressiveness in real time, applying high authority during large transients and reducing gains near convergence to limit control effort. Simulation results using a high-fidelity nonlinear quadcopter model demonstrate accurate trajectory tracking, bounded attitude excursions, smooth transition to hover after the final time, and consistent reward improvement, validating the effectiveness and robustness of the proposed learning-based gain scheduling strategy.

Keywords: Autonomy, Control Architectures, UAS Applications
Abstract: This paper presents a learning-based optimal control framework for stable and perception-aware quadrotor flight in near-ground, obstacle-adjacent, and close-proximity multi-robot environments central to Active Information Acquisition (AIA) tasks. In these regimes, quadrotors experience poorly modeled aerodynamic disturbances from propeller downwash and inter-vehicle interactions that degrade tracking and stability. We develop a Pontryagin-based Nonlinear Model Predictive Control (PMP-NMPC) framework that integrates optimal trajectory planning with online disturbance compensation. A spectrally normalized Deep Neural Network (DNN) learns residual downwash and interaction-induced dynamics and is embedded within the control loop to preserve boundedness and stability. A Lyapunov analysis establishes local finite-time practical stability under bounded disturbance approximation error. Simulations on Gym-PyBullet, a physics-based rigid-body simulation environment, demonstrate stable flight, effective disturbance rejection, and improved perception performance in disturbance-rich close-proximity scenarios.

Keywords: Control Architectures, UAS Applications, Manned/Unmanned Aviation
Abstract: This paper addresses the control of a quadrotor operating in a hybrid train–drone delivery system, focusing on payload uncertainty and landing on a moving railway platform. The considered mission includes take-off from a train entering a logistics terminal, cruise flight to a delivery location, and dynamic re-landing on the departing train. A hierarchical control architecture based on Model Predictive Control (MPC) is presented. It integrates trajectory generation, constrained MPC trajectory tracking, and an online parameter adaptation scheme based on recursive least squares for mass estimation. The adaptive model is embedded in the MPC prediction model, allowing real-time compensation of payload variations without direct payload measurements. Dynamic landing is formulated as a time-varying constrained tracking problem where the landing target evolves according to the train kinematic model. Simulation results validate the proposed architecture across the mission phases, showing accurate trajectory tracking, robustness to payload mismatch, and reliable landing on a moving platform.

Keywords: Control Architectures, Micro- and Mini- UAS
Abstract: Expanding the operational capabilities of Micro Air Vehicles (MAVs) hinges on control systems that manage highly nonlinear dynamics across broad flight envelopes. Incremental Nonlinear Dynamic Inversion (INDI) is popular for its simplicity and modest modeling needs, but its assumption of infinitely fast actuators and neglect of state-dependent effects limit performance when actuators have slow or heterogeneous dynamics or when aerodynamic effects are significant. Actuator Nonlinear Dynamic Inversion (ANDI) overcomes these limitations by explicitly incorporating state-dependent dynamics and finite actuator bandwidth into the control law, enabling improved tracking performance across diverse actuator configurations. This work implements the full ANDI stabilization controller on the Cyclone, a hybrid MAV tail-sitter, using cascaded complementary filtering for state estimation. Simulation and flight experiments validate the approach and assess whether this compensation yields practical performance gains, establishing ANDI as a viable, generic control solution for MAVs. Code is available at https://github.com/tudelft/paparazzi/

Keywords: Multirotor Design and Control, Aerial Robotic Manipulation, Control Architectures
Abstract: We present a robotics-oriented, coordinate-free formulation of inverse flight dynamics for fixed-wing aircraft on SO(3). Translational force balance is written in the world frame and rotational dynamics in the body frame; aerodynamic directions (drag, lift, side) are defined geometrically, avoiding local attitude coordinates. Enforcing coordinated flight (no sideslip), we derive a closed-form trajectory-to-input map yielding the attitude, angular velocity, and thrust–angle-of-attack pair, and we recover the aerodynamic moment coefficients component-wise. Applying such a map to tethered flight on spherical parallels, we obtain analytic expressions for the required bank angle and identify a specific zero-bank locus where the tether tension exactly balances centrifugal effects, highlighting the decoupling between aerodynamic coordination and the apparent gravity vector. Under a simple lift/drag law, the minimal-thrust angle of attack admits a closed form. These pointwise quasi‑steady inversion solutions become steady-flight trim when the trajectory and rotational dynamics are time-invariant. The framework bridges inverse simulation in aeronautics with geometric modeling in robotics, providing a rigorous building block for trajectory design and feasibility checks.

Keywords: Multirotor Design and Control, Control Architectures, Payloads
Abstract: Strict power limits on small drone interceptors constrain the use of reusable directed energy payloads such as RF jammers and High Power Microwave (HPM) sources. High-gain beams can close the engagement link budget with minimal transmit power, but their required tight Field of View (FoV) conflicts with the aggressive maneuvers of a pursuing multi-rotor. This paper resolves this conflict with a hardware-aware safety filter using Control Barrier Functions (CBFs) on the SE(3) manifold. The filter provides formal pointing guarantees, enabling highly directional antennas whose gain would otherwise be sacrificed for pointing tolerance. We formulate a phased array’s electronic steering limits as a relative-degree-two safety constraint with analytically verified Lie derivatives and solve the resulting Quadratic Program (QP) via ADMM with a constant-time core factorization. Validation in a nonlinear SE(3) simulation and a custom 6-DOF Software- in-the-Loop (SITL) environment demonstrates the proposed CBF eliminates all FoV violations and recovers up to 12.0 dB of antenna gain by tightening the beam. The approach also reveals conventional pitch clamping provides negligible benefit against FoV violations, avoids the severe actuator-saturation failure modes of multi-step Model Predictive Control (MPC)- CBF architectures, and executes reliably in under 350 µs.

Keywords: Multirotor Design and Control, Risk Analysis, Reliability of UAS
Abstract: This paper introduces Probabilistic Attainable Moment Sets for uncertainty-aware aircraft design optimization as an extension of the classical Attainable Moment Set concept to account for parametric uncertainty in actuator effectiveness. A computational method based on first-order parametric linearization of actuator models is proposed to propagate uncertainty through the control effectiveness mapping. The resulting probabilistic formulation enables the evaluation of control authority in terms of the likelihood of moment-space constraint violations rather than deterministic feasibility. Building on this framework, design optimization problems are formulated using a probabilistic metric that quantifies the probability of violating prescribed moment-space constraints. Two multirotor case studies are presented in which rotor tilt angles are optimized under actuator uncertainty, demonstrating the effectiveness of the proposed approach to uncertainty-aware aircraft design optimization.

Keywords: Multirotor Design and Control, Control Architectures, Technology Challenges
Abstract: Tilting multirotor platforms have gained significant attention in the research community due to their over-actuation and input redundancy capabilities. However, these features introduce complex challenges, such as the design of new nonlinear controllers and the use of control input allocation to fully exploit the redundancy of these platforms. A cascaded architecture comprising a high-level controller, a control allocation layer, and a low-level controller to command the actuators and rotors, offers great flexibility and space for co-design. This allows ad-hoc control allocation approaches to be defined without requiring modifications to the high-level controller. This review analyzes different methods of allocating control inputs presented in the literature and discusses their respective advantages and disadvantages. Furthermore, it categorizes these methods into three distinct classes based on their fundamental characteristics.

Keywords: Multirotor Design and Control, Simulation
Abstract: This paper proposes a new hybrid (i.e., physics-based and data-driven) modeling approach for multirotor UAVs. The model favors the physics component to promote simplicity, physical consistency, and suitability for control and estimation tasks. It relies on a physics-based dynamic formulation derived from Blade Element Theory and Momentum Theory, while restricting learning to the induced velocity of each rotor. We study the impact of rotor flow interactions by comparing models with and without interaction terms. We also propose a velocity-based learning procedure and compare it with a force-based alternative. The proposed approach is evaluated on a real world dataset. Results show that significant improvements in velocity prediction are obtained when rotor flow interactions are included. In addition, velocity-based training yields more accurate velocity prediction than force-based learning.

Keywords: Multirotor Design and Control, Air Vehicle Operations, UAS Applications
Abstract: In this paper, we present a generalized, comprehensive nonlinear mathematical model and conceptual design for the MetaMorpher, a metamorphic Unmanned Aerial Vehicle (UAV) designed to bridge the gap between vertical takeoff and landing agility and fixed-wing cruising efficiency. Building on the successful design of the spincopter platform, this work introduces a simplified mechanical architecture using lightweight materials and a novel wing-folding strategy. Unlike traditional rigid-body approximations, we derive a nonlinear flight dynamics model that enables arbitrary force distributions across a segmented wing structure. This modularity allows for testing different airfoils, mass distributions, and chord lengths in a single environment. As part of this work, various flight modes were specifically tested and analyzed in the Simulink environment. The results show that the model behaves predictably under different structural configurations, demonstrating its reliability as a tool for rapid design evaluation.

Keywords: Autonomy, UAS Testbeds, Swarms
Abstract: Unmanned aerial vehicles are rapidly transforming multiple applications, from agricultural and infrastructure monitoring to logistics and defense. Introducing greater autonomy to these systems can simultaneously make them more effective as well as reliable. Thus, the ability to rapidly engineer and deploy autonomous aerial systems has become of strategic importance. In the 2010s, a combination of high-performance compute, data, and open-source software led to the current deep learning and AI boom, unlocking decades of prior theoretical work. Robotics is on the cusp of a similar transformation. However, physical AI faces unique hurdles, often combined under the umbrella term ``simulation-to-reality gap''. These span from modeling shortcomings to the complexity of vertically integrating the highly heterogeneous hardware and software systems typically found in field robots. To address the latter, we introduce aerial-autonomy-stack, an open-source, end-to-end framework designed to streamline the pipeline from (GPU-accelerated) perception to (flight controller-based) action. Our stack allows the development of aerial autonomy using ROS2 and provides a common interface for two of the most popular autopilots: PX4 and ArduPilot. We show that it supports over 20x faster-than-real-time, end-to-end simulation of a complete development and deployment stack---including edge compute and networking---significantly compressing the build-test-release cycle of perception-based autonomy.

Keywords: UAS Testbeds, Smart Sensors, Technology Challenges
Abstract: The characterization of UAV propulsion systems requires reliable measurements of thrust and electrical performance, which are not always available from manufacturer datasheets under practical operating conditions. This paper presents the development and experimental validation of a low-cost instrumented static test bench for brushless motors used in UAV propulsion systems. The platform measures thrust, rotational speed, voltage, and current, enabling the analysis of propulsion performance and electrical power consumption. The mechanical structure was designed in CAD and fabricated using 3D printing with linear guides to reduce friction during thrust measurement. Sensor data are acquired using an Arduino- based system and processed in MATLAB for real-time visu- alization and analysis. Experimental validation was conducted using a Holybro motor–propeller assembly with reference data provided by the manufacturer. The results showed strong agreement with the expected propulsion behavior, confirming the quadratic relationship between thrust and RPM and the cubic trend between current and rotational speed. The proposed platform provides a reliable and accessible tool for propulsion testing and comparison of UAV motor–propeller configurations.

Keywords: UAS Testbeds, Swarms, Simulation
Abstract: Crisis and disaster response increasingly depends on persistent situational awareness over fast-changing hazards such as wildfires and floods. Fixed-wing UAV swarms are a promising enabler due to their wide-area coverage and endurance. Evaluating swarm mission supervision in these settings requires closed-loop scenarios that couple evolving environments to realistic vehicle dynamics and multi-agent execution. We present an integrated testbed for fixed-wing swarm supervision architectures that combine centralized mission reasoning with decentralized onboard behavior execution. The testbed integrates (i) a dynamic disaster scenario generator producing time-evolving hazard state and mission context, (ii) a multi-aircraft JSBSim-based flight dynamics simulation with control and energy modeling, and (iii) a decentralized guidance layer that executes a library of behavior primitives through a common interface and multi-rate updates. The framework supports deterministic seeding, synchronized replay, and logging of mission effectiveness and safety metrics for controlled comparisons. We validate the executability of primitives under calm and windy conditions within a bounded flight envelope, and demonstrate end-to-end closed-loop operation on generated scenarios. An LLM-based supervisor is used as an example client to exercise the mission interface and behavior primitives.

Keywords: UAS Testbeds, Simulation, Networked Swarms
Abstract: Maintaining network connectivity while managing limited battery resources is a real challenge in missions with multiple Unmanned Aerial Vehicles operating in cluttered environments. Although numerous connectivity-aware and energy-aware control strategies exist, they are often evaluated independently, limiting insight into their coupled effect. This paper presents a sim2real benchmarking framework for multi-robot UAV systems that enables systematic evaluation of connectivity preservation, battery management, and search mission performance under unified constraints. The framework integrates a Software-in-the-Loop Crazyflie simulator, a procedural world-generation pipeline, and a ROS 2-based interface and evaluation stack to ensure consistent deployment in both simulation and physical experiments. Communication constraints are modeled using range and line-of-sight–based connectivity graphs, while energy limitations require periodic returns to a charging base. Metrics quantify connectivity loss, recharging events, mission duration, and detection performance. In addition to that, several examples of benchmark worlds are provided. The platform was validated through deployment in the UAV Competition as a part of the 2025 International Conference on Unmanned Aircraft Systems, featuring simulation-based qualification and real-world finals. Experimental insights highlight remaining sim2real gaps in battery dynamics, communication robustness, and middleware configuration.

Keywords: UAS Testbeds, Autonomy, Simulation
Abstract: ROScopter is a lean multirotor autopilot built for researchers. ROScopter seeks to accelerate simulation and hardware testing of research code with an architecture that is both easy to understand and simple to modify. ROScopter is designed to interface with ROSflight 2.0 and runs entirely on an onboard flight computer, leveraging the features of ROS 2 to improve modularity. This work describes the architecture of ROScopter and how it can be used to test application code in both simulated and hardware environments. Hardware results of the default ROScopter behavior are presented, showing that ROScopter achieves similar performance to another state-of-the-art autopilot for basic waypoint-following maneuvers, but with a significantly reduced and more modular code-base.

Keywords: UAS Testbeds, Simulation, Autonomy
Abstract: ROSflight is a lean, open-source autopilot ecosystem for unmanned aerial vehicles (UAVs). Designed by researchers for researchers, it is built to lower the barrier to entry to UAV research and accelerate the transition from simulation to hardware experiments by maintaining a lean (not full-featured), well-documented, and modular codebase. This publication builds on previous treatments and describes significant additions to the architecture that improve the modularity and usability of ROSflight, including the transition from ROS 1 to ROS 2, supported hardware, low-level actuator mixing, and the simulation environment. We believe that these changes improve the usability of ROSflight and enable ROSflight to accelerate research in areas like advanced-air mobility. Hardware results are provided, showing that ROSflight is able to control a multirotor over a serial connection at 400 Hz while closing all control loops on the companion computer.

Keywords: UAS Applications
Abstract: The secure landing of an Unmanned Aerial Vehicle (UAV) in a cluttered environment is considered. Using a downward-facing depth sensor, the UAV detects obstacles by taking depth point-cloud measurements. These measurements are rectified depending on the UAV's roll and pitch angles provided by its autopilot. The planes of the obstacles are then extracted from the normal vectors using an agglomerative clustering technique. Planes with normal directions perpendicular to the gravity vector are subsequently excluded. The remaining planes are feasible candidates for landing. Using a UAV-centered circular convex hull, areas within those non-convex planes are sought within which the UAV can safely land. Using a variation of the poles of inaccessibility algorithm, the maximum area-wise inscribed circle is computed in each plane. If more than one circle is computed containing the UAV's hull, the farthest one is selected depending on the average depth value. Experimental studies using a quadrotor are offered to highlight the efficiency of this real-time landing scheme.

Keywords: UAS Applications, Fail-Safe Systems, Reliability of UAS
Abstract: Autonomous multirotor landings on uncrewed surface vessels (USVs) are critical for persistent maritime operations but remain challenging due to wave-induced tilt, wind disturbances, and limited landing area. Many existing approaches exhibit small pose tolerance for reliable landing. This paper presents a lightweight toggleable adhesion mechanism to improve landing reliability. The system uses a motor-driven corkscrew that engages hook-and-loop material on the landing surface, enabling active adhesion during landing and controlled release during takeoff. We evaluate a prototype using a modified Crazyflie 2.0 and a custom tilting platform at fixed angles representative of extreme wave conditions. Using only a simple vertical PID controller, the proposed approach increases landing success from an average of 40% (baseline) to 80% across platform tilts up to 43 degrees using appropriately selected actuation settings.

Keywords: UAS Applications, Micro- and Mini- UAS, Control Architectures
Abstract: Efficient autonomous exploration in unknown, obstacle-cluttered environments remains a significant challenge in the robotic field. Building on our previous Partial Differential Equation-based navigation for ground vehicles cite{Kotsinis2025}, this paper extends the framework to Micro Aerial Vehicles operating under severe sensing constraints. Using a Crazyflie 2.1 quadrotor equipped with a sparse four-beam range sensor, we propose a continuous exploration policy that simultaneously optimizes translation and rotation to maximize information gain. We generate these velocity commands by solving an elliptic Partial Differential Equation with Dirichlet boundary conditions. To overcome flat gradient deadlocks, we solve a Poisson equation with source values dynamically derived from a Hybrid Visibility Graph. Crucially, we enhance the grid-free Walk on Spheres algorithm by reducing solution variance, significantly lowering computational overhead. Validated through comparative simulations and real-world ROS2-MATLAB experiments, our approach efficiently calculates the navigation gradient directly at the agent's position, ensuring smooth, deadlock-free exploration in complex 2D spaces.

Keywords: UAS Testbeds, Perception and Cognition, UAS Applications
Abstract: This work describes the design features and systems of the Laysan, a small UAS with VTOL capability, waterproof environmental resilience, and solar energy-harvesting with dynamic soaring augmentation for migratory cross-ocean missions. Key innovations include a radar-based altimetry system, a self-cleaning pitot-static airspeed sensing system, and a unique camera-based meteorological perception system running in a real-time on-board Neural Processing Unit. The radar altimeter and airspeed are used for dynamic soaring maneuvering, and the outputs from the perception system can be leveraged for path-planning avoidance of clouds and other hazardous weather. Experiments with these systems are demonstrated, such as the behavior of the radar altimeter mounted to the aircraft in a real-world over-water dynamic soaring cycle, and the performance of the trained cloud instance segmentation system.

Keywords: Swarms, UAS Applications, Navigation
Abstract: This article presents a solution to intercept an agile drone by a team of agile drone carrying catching nets. We formulate the problem as a competitive Multi-Agent Reinforcement Learning (MARL) task. To address the problem of non-stationarity and catastrophic forgetting of agents overfitting to the current opponent strategy, we train the pursuers and the evader using Multi-Agent Proximal Policy Optimization (MAPPO) with Prioritized Fictitious Self Play (PFSP). We train the agents in a high-fidelity simulator using low-level control commands, collective thrust and body rates (CTBR), to achieve agile flights for both the pursuers and the evader. We compare the performance of the trained policies in terms of catch rate, time to catch and crash rates, against heuristic baselines and show that our solution outperforms them. Ablation studies show that PFSP lead to more robust policies that can adapt to different opponent strategies, and that a low-level control commands are crucial for learning performing strategies in the pursuit-evasion task. Finally, a qualitative analysis of the learned behaviours highlights the emergence of cooperative tactics among the pursuers.

Keywords: Micro- and Mini- UAS, Multirotor Design and Control, Navigation
Abstract: This paper presents the modeling, identification, and experimental control of a Parrot Mambo micro–Unmanned Aerial Vehicle (UAV) using a Linear Quadratic Gaussian (LQG) control framework. A dynamic model of the quadrotor is developed by combining physical inspection with experimental system identification to estimate key parameters such as inertial properties, thrust coefficients, and geometric characteristics. The resulting model is expressed in state-space form and used to design an optimal state-feedback controller coupled with Kalman filtering for state estimation. The control architecture follows a cascade structure that separates position and attitude control loops, enabling stable and reliable trajectory tracking under sensor noise and modeling uncertainties. Controller performance is evaluated through both simulation and real flight experiments using representative square and circular trajectories. Experimental results show accurate trajectory tracking and improved transient behavior compared with a conventional PID baseline. In addition, the study provides an experimentally validated dynamic model of the Parrot Mambo platform, highlighting the importance of parameter identification and state estimation in achieving reliable control of micro-UAV systems.