We present a compositional approach to the early modeling and analysis of complex aerospace systems based on assume–guarantee contracts. Components in a system are abstracted into assume–guarantee specifications. Performing algebraic contract operations with Pacti allows us to relate local component specifications to those of the system. Applications to two aerospace case studies—the design of spacecraft to satisfy a rendezvous mission and the design of the thermal management system of a prototypical aircraft—show that this methodology provides engineers with an agile, early analysis and exploration process. We show that the analysis of thousands of configurations can be achieved on the order of seconds to minutes on a consumer-grade CPU.
This paper presents a comparative case study of wildland fire rate of spread (ROS) estimation, contrasting infrared data from an autonomous unmanned aerial system (UAS) with predictions from the Rothermel fire behavior model during a prescribed burn conducted in Zaleski State Forest, Ohio. In this study, we built and deployed a small UAS platform equipped with infrared sensing to measure the ROS of the fire front in a forest with significant topographic variation. Infrared data from the experiment were reprojected into the three-dimensional world frame, followed by a geometrical analysis of the observed fire intensity contours. Delaunay triangulation was performed to discretize the space around time-separated fire fronts, resulting in local ROS estimates based on fire intensity gradients. The infrared-derived ROS estimates were then compared with widely used Rothermel model predictions. Finally, a vector alignment analysis was conducted to investigate discrepancies between the methods by comparing ROS directionality with local terrain gradients. The results showed a small to negligible correlation, suggesting that the observed fire-front motion was not strongly aligned with terrain-gradient direction alone. These results underscore the complexities of fire behavior shaped by terrain, environmental factors, and the limitations of predictive models in natural fire conditions.
Space systems generate vast quantities of multivariate time-series data from numerous sensors, reflecting the complex interlinking of physical parameters within a dynamic system. As the volume, complexity, dynamism, and multivariance of this data increase, conventional anomaly detection methods become suboptimal for accurately identifying and labeling outliers. In physics-guided approaches, physics is typically built into the loss function as a single equation, or the data are generated through simulation of a physical system. The novelty of this framework, however, lies in coding all equations governing the system as relationships between parameters. These relationships are captured through a multimetric approach and encoded into an image. Subsequently, interpretability for anomaly detection is achieved through dynamic relationship graphs drawn from this image. The strength of this physics-informed, zero-shot training framework is demonstrated by its ability to identify anomalous regions, classifying systems based on physical properties, and it also serves as a diagnostic framework for the digital twin of on-orbit systems. Indeed, this concept can be applied to any complex scientific system to highlight violations of scientific laws, whether they stem from sensor failure, subsystem failure, or system failure. Furthermore, it has been validated using Tennessee Eastman Process data, providing confidence in its effectiveness.
This research addresses the need to develop in-space refueling technologies by experimentally demonstrating an autonomous docking maneuver and fluid transfer operation between a servicer spacecraft and an uncooperative spinning client spacecraft. A novel computer vision algorithm estimates the relative pose of the client spacecraft in real time using an onboard light detection and ranging camera. An unscented Kalman filter refines the pose data before a model predictive controller generates a collision-free docking trajectory for the servicer spacecraft to follow. A custom-designed docking and refueling mechanism for fluid transfer performs magnetic soft docking and robotic arm capture to seal a probe-and-drogue connection and transfer fluid from the servicer to the client spacecraft. Planar experimental validation is conducted at Carleton University’s Spacecraft Proximity Operations Testbed, demonstrating a complete autonomous docking and refueling operation.
With characteristics such as low cost, flexibility, and scalability, unmanned aerial vehicle (UAV) swarms demonstrate superior performance over single unmanned platforms and manned aircraft in search and surveillance missions. However, the conflicts between individual decisions and the tradeoff between search and connectivity in limited sensing areas still render UAV swarm search and surveillance inherently challenging. This paper proposes a subgroup differentiation-based decision-making framework for the UAV swarm, where two kinds of roles (informers and relays) are considered and each UAV can autonomously switch its role according to the task demand and environmental changes. The relay nodes provide larger communication scopes for connectivity preservation and contribute to the relaxation of informers’ constrained actions. In this way, the informing UAVs can maintain well-preserved explorations during the search process. The subgroup differentiation is based on a distributed framework where two sequentially linked auction-based operators are respectively developed for the action policies of informers and relays. The impact time control guidance is used for simultaneous arrival to realize the synchronous information fusion of swarm search findings. Simulation results demonstrate the efficient explorations, less conservativeness, and higher coverage efficiencies of the proposed method over existing advanced approaches in confined sensing environments.
Surrogate-based optimization using neural networks (NNs) reduces computational costs in engineering design but depends heavily on systematic hyperparameter optimization (HPO). This study compares HPO methods-including grid search, random search, Bayesian optimization (BO), hyperband (HB), and BOHB-using analytical functions and aerodynamic shape optimization (ASO). The study first explores HPO in a one-shot method where NNs are trained with datasets of different sample sizes. The analysis then extends to adaptive HPO approaches within an efficient global optimization (EGO) framework using NNs, which employs sequential sampling. Under this framework, static HPO (maintaining initially optimized HPs), periodic HPO (adjusting HPs every five infill points), and dynamic HPO (adjusting HPs after each infill point) are compared. In ASO with a one-shot method, BOHB achieves a drag coefficient (Cd) of 117 drag counts (d.c.)-close to BO's 115 d.c.-while requiring only 30.5% of BO's computational time with 500 samples. Additionally, in ASO with sequential sampling, periodic HPO effectively balances performance and computational efficiency by achieving a Cd of 119 d.c. with 25 initial samples and 50 infills. Dynamic HPO reduces Cd to 113 d.c. but at a higher cost compared to periodic HPO, which offers a balance between drag reduction and computational expense.
Airport delay prediction is a key technological support for proactive intervention in the air transportation system, enhancing decision-making for airlines, air traffic control authorities, and airport operators. Accurate forecasts optimize resource allocation and alleviate delay impacts. However, the complex spatiotemporal nature of airport networks and delay propagation poses significant challenges, as traditional statistical, machine learning, and single-series forecasting approaches often fail to capture intricate dependencies. To address this, we propose a graph convolutional network (GCN)-transformer-gated recurrent unit (GRU) model. First, an adjacency matrix derived from historical delays encodes spatial relationships, which are extracted by the GCN. Then, a transformer encoder models global temporal dependencies via self-attention. Finally, the GRU integrates multi-source inputs such as weather data for multi-step decoding to predict delays across future intervals. Case studies on China's 2023 airport operation dataset show that our model outperforms random forest, long short-term memory, standalone GRU, spatiotemporal GCN, and attention-based spatiotemporal GCN models, yielding lower prediction errors and better model fit. Incorporating delay-based adjacency and weather factors further enhances performance. These results demonstrate a clear advantage in prediction accuracy and decision support.
The complexity of aerospace systems often leads to the use of multiple computer simulation models. These can support decision-making but are difficult to integrate because of their heterogeneity. In addition, most modeling efforts focus on system design instead of operations. Calibrating multiple models uniformly is difficult, as uncertainties must be aggregated and quantified. We introduce a method based on Bayesian networks capturing multimodel dependencies to infer model parameters. The method automates calibration for decision-makers to conduct “what-if” simulations on unanticipated operational scenarios. There is little literature studying how Bayesian networks should be designed to address multimodel calibration as, usually, such networks are used to represent real-world entities rather than model parameters. We address this gap by introducing a design process that uses model-based systems engineering resources and comparing it with a baseline process. These processes are demonstrated in a deep space exploration habitat study. We found that network construction can benefit from 1) defining utility metrics that quantify the operational value of parameters, 2) delaying computational cost reduction efforts, and 3) combining topology-driven and causality-driven architectures when defining the graph structure.
Physical or operational constraints often impose communication limitations on autonomous agents. Such limitations complicate monitoring or multi-agent coordination. Even when strong communications are absent, some information may still be available. The remainder of the relevant agent state may be reconstructed via estimation. The actions taken by an agent are a potential source of information-as the agent interacts with the environment, these actions may be observed even in the absence of explicit communication. We investigate using actions to estimate the state of an agent, using reinforcement learning to develop policies that make the estimation problem more tractable. Policy observability is encouraged through the training reward and is analyzed using simulations of the trained agent. In an aircraft tracking problem, a policy with enhanced observability is found that has minimal impact on nominal task performance.
Runway overrun is a major threat to civil aviation safety. Many studies of runway overruns have been conducted on risk assessment, risk prediction, and risk factor identification, but data-driven quantitative operational guidance for pilots remains limited in the literature. Airspeed exceedance in the landing phase has been identified as one of the major risk factors of long landing and runway overruns. By taking advantage of data collected from the Quick Access Recorder, this paper proposes a novel method utilizing the Kolmogorov-Smirnov index to identify the airspeed safety envelope, which is defined as a monotonic, smooth function of "airspeed safety threshold versus height" during the landing phase. Analysis results show that the probability of long landing significantly increases if the airspeed exceeds the identified envelope, and pilots are suggested to keep the airspeed below the safety envelope to reduce the risk of long landing. The proposed framework is easy to implement and interpret, with an efficient algorithm. It provides a starting point for similar applications aimed at identifying safety thresholds and envelopes of key operational parameters and providing practical guidance for pilots.
This paper presents a physics-based data-driven approach for predicting the remaining useful life (RUL) of a spacecraft reaction wheel (RW). Our method combines a physics-based model with a data-driven regression and machine learning technique known as the Sparse Identification of Nonlinear Dynamics Systems with Control Input (SINDYc). This approach is used for fault detection and RUL prediction of the RW. For fault detection, we predict the states and health index (HI) parameters of the RW, with the coefficients of output torque and viscous friction selected as the HI parameters. To estimate the RUL, we analyze the trends of these HI parameters over time, predicting when the failure threshold will be crossed. We demonstrate that the proposed method is more effective and suitable for autonomous onboard applications compared to existing methods, such as long short-term memory recurrent neural networks. SINDYc also provides explicit identification of the underlying RW governing equations while requiring orders of magnitude less computational effort to produce a model. We establish the robustness of this methodology by confirming only slight degradation in model accuracy when fitting to noisy data.
Colombia currently lacks sovereign space situational awareness (SSA) capabilities despite the growing strategic importance of protecting national space assets such as the FACSAT constellation. This paper proposes a hybrid radar-enhanced SSA architecture tailored to Colombia's geographical, climatic, and institutional constraints, integrating dual-frequency (L/X-band) ground radars, CubeSat optical assets, and an AI-enabled multisensor fusion framework. The methodology combines analytical propagation modeling, environmental attenuation assessment, sensor-geometry analysis, and a fusion pipeline incorporating classical filtering and machine-learning-based residual estimation. Simulation results indicate that the proposed architecture can achieve high levels of LEO detection coverage-approaching 90-95% under idealized sensor geometries-and low-latency orbital updates, with performance dependent on radar siting, environmental conditions, and the deployment of edge computing nodes. The study provides an evidence-based roadmap for Colombia to develop scalable, weather-resilient SSA capabilities and reduce reliance on foreign networks. The findings highlight the feasibility of a phased implementation strategy that strengthens national autonomy while enabling future regional cooperation in Latin American space safety and surveillance.
Rapidly expanding low Earth orbit satellite constellations are placing increasing demands on terrestrial ground networks, motivating the development of more efficient ground station network designs. Current approaches select sites from predefined locations, limiting optimization to existing infrastructure and constraining performance. In contrast, free-placement optimization operates over a continuous spatial domain on Earth, broadening the search space and allowing higher-throughput configurations at the cost of potentially requiring new infrastructure deployment. In this work, we introduce SCORE (Sequential Cyclic Optimization via Refinement Evaluation), a two-stage free-placement method for ground station design. SCORE combines sequential coordinate selection with cyclic refinement to manage high-dimensionality, non-convexity, and local minima that challenge global optimizers. We benchmark SCORE against one-shot methods such as differential evolution (DE) and integer programming approaches using locations from Kongsberg Satellite Services and the World Teleport Association. Tests across two commercial Earth observation constellations (Capella Space and ICEYE) and one synthetic Walker-Star constellation show that SCORE requires up to 5x fewer function evaluations to converge relative to DE while improving downlink throughput by up to 13
This paper presents a novel framework for optimizing the locations and sizes of urban air mobility (UAM) vertiports. Unlike previous studies that focused on maximizing accessibility and potential demand, our approach emphasizes the social acceptance of UAM. A vertiport location problem with social benefits (VLPSB) that maximizes social benefits while considering practical constraints such as demand and noise pollution is proposed. Additionally, we introduce a stepwise vertiport location optimization problem (S-VLPSB) to address the demand uncertainty in UAM by solving the VLPSB sequentially. A case study and a comparative study for Jeju Island in South Korea, a representative scenario based on actual data, demonstrate the validity of the framework and discuss the impact and implications of key parameter variations.
