Organizational Unit:

Daniel Guggenheim School of Aerospace Engineering

Permanent Link

https://hdl.handle.net/1853/70742

Parent Organization

Organizational Unit

College of Engineering

Includes Organization(s)

Organizational Unit

Aerospace Design Group

Organizational Unit

Aerospace Systems Design Laboratory (ASDL)

Organizational Unit

Air Transportation Laboratory

Organizational Unit

Cognitive Engineering Center

Organizational Unit

Space Systems Design Laboratory (SSDL)

Show 1 more

ArchiveSpace Name Record

https://finding-aids.library.gatech.edu/agents/corporate_entities/106

Full item page

Publication Search Results

Now showing 1 - 10 of 302

Runway safety improvements through a data driven approach for risk flight prediction and simulation

(Georgia Institute of Technology, 2022-12-19) Lee, Hyunki

Runway overrun is one of the most frequently occurring flight accident types threatening the safety of aviation. Sensors have been improved with recent technological advancements and allow data collection during flights. The recorded data helps to better identify the characteristics of runway overruns. The improved technological capabilities and the growing air traffic led to increased momentum for reducing flight risk using artificial intelligence. Discussions on incorporating artificial intelligence to enhance flight safety are timely and critical. Using artificial intelligence, we may be able to develop the tools we need to better identify runway overrun risk and increase awareness of runway overruns. This work seeks to increase attitude, skill, and knowledge (ASK) of runway overrun risks by predicting the flight states near touchdown and simulating the flight exposed to runway overrun precursors. To achieve this, the methodology develops a prediction model and a simulation model. During the flight training process, the prediction model is used in flight to identify potential risks and the simulation model is used post-flight to review the flight behavior. The prediction model identifies potential risks by predicting flight parameters that best characterize the landing performance during the final approach phase. The predicted flight parameters are used to alert the pilots for any runway overrun precursors that may pose a threat. The predictions and alerts are made when thresholds of various flight parameters are exceeded. The flight simulation model simulates the final approach trajectory with an emphasis on capturing the effect wind has on the aircraft. The focus is on the wind since the wind is a relatively significant factor during the final approach; typically, the aircraft is stabilized during the final approach. The flight simulation is used to quickly assess the differences between fight patterns that have triggered overrun precursors and normal flights with no abnormalities. The differences are crucial in learning how to mitigate adverse flight conditions. Both of the models are created with neural network models. The main challenges of developing a neural network model are the unique assignment of each model design space and the size of a model design space. A model design space is unique to each problem and cannot accommodate multiple problems. A model design space can also be significantly large depending on the depth of the model. Therefore, a hyperparameter optimization algorithm is investigated and used to design the data and model structures to best characterize the aircraft behavior during the final approach. A series of experiments are performed to observe how the model accuracy change with different data pre-processing methods for the prediction model and different neural network models for the simulation model. The data pre-processing methods include indexing the data by different frequencies, by different window sizes, and data clustering. The neural network models include simple Recurrent Neural Networks, Gated Recurrent Units, Long Short Term Memory, and Neural Network Autoregressive with Exogenous Input. Another series of experiments are performed to evaluate the robustness of these models to adverse wind and flare. This is because different wind conditions and flares represent controls that the models need to map to the predicted flight states. The most robust models are then used to identify significant features for the prediction model and the feasible control space for the simulation model. The outcomes of the most robust models are also mapped to the required landing distance metric so that the results of the prediction and simulation are easily read. Then, the methodology is demonstrated with a sample flight exposed to an overrun precursor, and high approach speed, to show how the models can potentially increase attitude, skill, and knowledge of runway overrun risk. The main contribution of this work is on evaluating the accuracy and robustness of prediction and simulation models trained using Flight Operational Quality Assurance (FOQA) data. Unlike many studies that focused on optimizing the model structures to create the two models, this work optimized both data and model structures to ensure that the data well capture the dynamics of the aircraft it represents. To achieve this, this work introduced a hybrid genetic algorithm that combines the benefits of conventional and quantum-inspired genetic algorithms to quickly converge to an optimal configuration while exploring the design space. With the optimized model, this work identified the data features, from the final approach, with a higher contribution to predicting airspeed, vertical speed, and pitch angle near touchdown. The top contributing features are altitude, angle of attack, core rpm, and air speeds. For both the prediction and the simulation models, this study goes through the impact of various data preprocessing methods on the accuracy of the two models. The results may help future studies identify the right data preprocessing methods for their work. Another contribution from this work is on evaluating how flight control and wind affect both the prediction and the simulation models. This is achieved by mapping the model accuracy at various levels of control surface deflection, wind speeds, and wind direction change. The results saw fairly consistent prediction and simulation accuracy at different levels of control surface deflection and wind conditions. This showed that the neural network-based models are effective in creating robust prediction and simulation models of aircraft during the final approach. The results also showed that data frequency has a significant impact on the prediction and simulation accuracy so it is important to have sufficient data to train the models in the condition that the models will be used. The final contribution of this work is on demonstrating how the prediction and the simulation models can be used to increase awareness of runway overrun.
A Methodology for the Design and Operational Safety Assessment of Unmanned Aerial Systems

(Georgia Institute of Technology, 2022-12-15) Kendall, Andrew Paul

Efforts are underway to introduce Unmanned Aerial Systems (UAS) into routine cargo operations within the National Airspace System (NAS). Such systems have the potential to increase transport system flexibility by mitigating crew scheduling constraints and extending operations to remote locations. It is expected that any large UAS operating in the transport category must comply with Federal Aviation Regulations to achieve airworthiness certification for routine operations within the NAS. Regulations on the safety of equipment, systems, and installations require all failure conditions due to malfunctions, environmental events, and inadequate corrective action to be mitigated and shown to be extremely improbable. These system safety requirements are particularly relevant for a UAS as the ability of a Remote Pilot (RP) to detect and respond to risks is dependent on a Command and Control (C2) link. Failure conditions associated with the C2 link system require autonomy onboard the aircraft to supplement the RP in order to mitigate risk. A method for assessing the performance required from automation when the RP cannot adequately mitigate risks is needed to allow routine UAS operations. The problem of ensuring autonomous UAS safety requirements is addressed in this thesis through the development of a safety assessment methodology that can be applied during both system design and online operations. The contributions are as follows: • Safety Regulations are formulated as a chance-constraint satisfaction problem, requiring safety on the order of 1 accident per billion operations. Rare event estimation techniques based on Importance Sampling are proposed to assess safety subject to various sources of uncertainty. • Failure conditions can be due to both discrete events, such as system failures, and continuous state uncertainties, such as navigation errors and turbulence. A stochastic hybrid system model is proposed to handle the coupling between discrete and continuous states and estimate the distribution of aircraft trajectories that may result from a given set of system parameters, operational conditions, and decision parameters. • The final approach and landing phase of flight serves as a use case for the methodology. The safety assessment is applied to determine system design parameters required to passively mitigate risks. The methodology is extended to active risk mitigation during operations, where online safety assessments using updated observations are used to ensure decision options always exist that will satisfy safety requirements.
Framework for the design and operations of sustainable on-orbit servicing infrastructures dedicated to geosynchronous satellites

(Georgia Institute of Technology, 2022-12-15) Sarton Du Jonchay, Tristan

After being a concept for decades, the on-orbit servicing industry is finally taking off, with national space agencies and private organizations developing and planning soon-to-be-launched space infrastructures that will revolutionize the way humans operate in space. The advent of this new industry comes at a time when the geosynchronous orbit (GEO) satellite industry faces various pressures whether it be because of ageing fleets or increased competition from nimbler Low Earth Orbit (LEO) and Medium Earth Orbit (MEO) constellations. A new symbiotic relationship is emerging between early OOS players who seek customers and the GEO satellite operators who aim to revive the competitiveness of their fleets. The first OOS infrastructures will be simple ones, involving a few servicers offering a narrow set of services. These servicers will provide services to a few satellites before running out of propellant and getting discarded in a graveyard orbit or into the atmosphere. However, as technology matures and demand for on-orbit services increases, OOS infrastructures will become more versatile and involve additional elements, such as orbital depots, to enable the sustainable operations of a wide variety of servicers. Thus, planning OOS missions will involve not only finding the best route for every single servicer but also optimizing the in-space supply chain of commodities needed to support the long-term operations of the servicers and their client satellites. This dissertation presents an OOS planning framework that simultaneously computes the optimal route of the servicers and plans the in-space supply chain of the supporting commodities. The second chapter gives the background of OOS in GEO and the literature review for OOS planning relevant to the work presented in this thesis. The third chapter presents the mission scenario investigated in this work. The fourth chapter generalizes the Time-Expanded Generalized Multi Commodity Network Flow (TE-GMCNF) model used in recent state-of-the-art space logistics studies to accurately model the operations of the servicers across a network of customer satellites and orbital depots. The Rolling Horizon (RH) approach is adapted to the OOS context to properly model uncertain service demand arising from customer satellites. The fifth chapter generalizes the mathematical formulation at the core of the framework developed in chapter 4 to model all kinds of user-defined trajectories and servicer propulsion technologies, such as high-thrust, low-thrust, and/or multimodal servicers. (Multimodal servicers are defined to be equipped with both high-thrust and low-thrust engines.) An assumption inherent to chapter 4 and chapter 5 is that the nodes of the networks are all co-located along the same orbit. Chapter 6 relaxes this assumption by extending the framework developed in chapter 4 through the computation of the relative dynamics of network nodes distributed across orbits of various shapes and orientations. Thus, chapter 6, unlike chapter 4 and chapter 5, optimizes the operations of OOS infrastructures over a network with time-varying arc costs.
Sampling-based Dynamic Optimization: Theory, Analysis and Applications

(Georgia Institute of Technology, 2022-12-14) Wang, Ziyi

Dynamic optimization solves for the optimal strategy for a system that evolves over time. It is an integral part of many disciplines such as optimal control theory and robotics. Sampling-based methods have gained much popularity in recent years for solving dynamic optimization problems due to their ability to handle discontinuities in the dynamics and cost function. While many sampling-based dynamic optimization algorithms are developed, they are based on different theoretical foundations and are often designed for specific applications with heuristics. This thesis aims to bridge the gap between different theories, derive the general forms of several state-of-the-art sampling-based dynamic optimization algorithms and their extensions to various problem formulations. We present three general perspectives for deriving sampling-based dynamic optimization algorithms, namely Variational Optimization, Variational Inference and Stochastic Search. We show the equivalence between Variational Optimization and optimal control theory under certain assumptions. We justify previously used heuristics with proper problem formulation and derivation. In addition, we demonstrate that state-of-the-art Model Predictive Path Integral and Cross Entropy Method algorithms can be recovered as special cases under these perspectives. We discuss the connections between the perspectives and the unique algorithmic characteristics of each perspective. A unifying analysis is performed on the convergence, sampling complexity and variance. Based on the three perspectives, we develop Model Predictive Control algorithms for several application scenarios. We first apply the different schemes to standard stochastic optimal control problems. A risk sensitive extension is also derived to optimize with respect to the conditional value-at-risk of the cost function. The resulting algorithm performs optimization on non-Gaussian beliefs provided by a particle filter. We also apply the Stochastic Search perspective to the complex jump diffusion process and opinion dynamics. Finally, we propose a general distributed framework for scaling sampling-based dynamic optimizers for large-scale multi-agent control using consensus Alternating Direction Method of Multipliers. The effectiveness and applicability of the proposed algorithms are highlighted by results on various systems from control theory and robotics in simulation. In particular, the distributed framework is tested on a 200-agent Dubins vehicle formation control task.
Hybrid Sensor Networks for Active Monitoring: Collaboration, Optimization, And Resilience

(Georgia Institute of Technology, 2022-12-14) Guo, Yanjie

Hybrid sensor networks (HSN) consist of both static and mobile sensors deployed to fulfill a common monitoring task. The hybrid structure generalizes the network’s design problem and offers a rich set of possibilities for a host of environmental monitoring and anomaly detection applications. HSN also raise a new set of research questions. Their deployment and optimization provide unique opportunities to improve the network’s monitoring performance and resilience. This thesis addresses three challenges associated with HSN related to the collaboration, optimization, and resilience aspects of the network. Broadly speaking, these challenges revolve around the following questions: (1) how to collaboratively allocate the static sensors and devise the path planning of the mobile sensors to improve the monitoring performance? (2) how to select and optimize the sensor portfolio (the mix of each type of sensors) under given cost constraints? And (3) how to embed resilience in a HSN to sustain the monitoring performance in the face of sensor failures and disruptions? In part I, collaboration, this thesis develops a novel deployment strategy for HSN. The strategy solves the static sensor allocation problem, the mobile sensor path planning problem, and most importantly, the collaboration between these two types of sensors. Previous research in this area has addressed these problems separately in simplified environments. In this thesis, a collaborative deployment strategy of HSN is developed to improve the ultimate monitoring performance in complex environments with obstacles and non-uniform risk distribution. In part II, optimization, this thesis addresses the HSN sensor portfolio selection problem. It investigates the tradeoff between the static and mobile sensors to achieve the optimal monitoring performance under different cost constraints. Previous research in this area has studied the optimization problem for networks with a single type of sensor. In this thesis, a general optimization problem is formulated for HSN with static and mobile sensors and solved to identify the optimal portfolio mix and its main drivers. In part III, resilience, this thesis identifies monitoring resilience as a key feature enabled by HSN. This part focuses on the performance degradation of HSN in the presence of sensor failures and disruptions, and it identifies the means to embed resilience in a HSN to mitigate this performance degradation. Monitoring resilience is achieved by accounting for potential sensor failures in the deployment strategy of both static and mobile sensors through a novel, carefully designed probability sum technique. Previous research in this area has examined the reliability problem from a coverage point of view. This thesis extends the scope of investigation of HSN from reliability to resilience, and it shifts the focus from coverage considerations to the actual monitoring performance (e.g., detection time lag) and its resilience in the face of disruptions. To demonstrate and validate this novel perspective on HSN and the associated technical developments, this thesis focused on two examples of fire detection in a multi-room apartment using temperature sensors and CO leak detection in a 3D space station module with ventilation system. Three metrics are adopted as the ultimate monitoring performance, namely the detection time lag, the anomaly source localization uncertainty, and the state estimation error. A simulation environment based on the advection-conduction heat propagation model is developed for the computational experiments. The results (1) demonstrate that the optimal collaborative deployment strategy allocates the static sensors at high-risk locations and directs the mobile sensors to patrol the rest of the low-risk areas; (2) identify a set of conditions under which HSN significantly outperform purely static and purely mobile sensor networks across the three performance metrics here considered; and (3) establish that while sensor failures can considerably degrade the monitoring performance of traditional static sensor networks, the resilient deployment of HSN drastically reduces the performance degradation.
The design, education and evolution of a robotic baby

(Georgia Institute of Technology, 2022-12-13) Zhu, Hanqing

Inspired by Alan Turing’s idea of a child machine, I introduce the formal definition of a robotic baby, an integrated system with minimal world knowledge at birth, capable of learning incrementally and interactively, and adapting to the world. Within the definition, fundamental capabilities and system characteristics of the robotic baby are identified and presented as the system-level requirements. As a minimal viable prototype, the Baby architecture is proposed with a systems engineering design approach to satisfy the system-level requirements, which has been verified and validated with simulations and experiments on a robotic system. The capabilities of the robotic baby are demonstrated in natural language acquisition and semantic parsing in English and Chinese, as well as in natural language grounding, natural language reinforcement learning, natural language programming and system introspection for explainability. Furthermore, the education and evolution of the robotic baby are illustrated with real-world robotic demonstrations. Inspired by the genetic inheritance in human beings, knowledge inheritance in robotic babies and its benefits regarding evolution are discussed.
Inter-scale energy transfer in turbulent premixed combustion

(Georgia Institute of Technology, 2022-12-07) Kazbekov, Askar

Turbulent premixed combustion is widely used for energy conversion in power generation and propulsion devices. However, our understanding of the underlying fluid dynamics, combustion, and their interaction is still incomplete. The complexity of turbulent combustion arises from the non-linear, multi-scale, and multi-physics nature of the problem, which involves interactions between fluid dynamic and chemical processes across a myriad of length and time scales. The existing literature demonstrates that the dynamics of reacting turbulence does not necessarily follow the same phenomenology as in non-reacting incompressible turbulence. One of the key differences in reacting and compressible flows is the reversal of the classical turbulent energy cascade in a process termed as ‘backscatter’. Moreover, backscatter was shown to potentially depend on the magnitude of the pressure gradients across the flame; this is reflected in the sub-filter-scale pressure-work. Previous studies have predominantly focused on flames in homogeneous isotropic turbulence (HIT), in which the pressure gradients are associated with the flame and turbulence themselves. In contrast, practical combustors have mean pressure fields generated by the flow, which can induce significantly different turbulence dynamics as compared to non-reacting turbulence. The presented research explores the conditions at which energy backscatter occurs in an aerospace relevant configuration and attempts to identify the underlying physical mechanisms that have a leading order impact on these processes. This is done through systematic variation of the global equivalence ratio, the jet flow velocity, and the swirl number. The impact of these controlling parameters on turbulence production and energy backscatter is assessed through the analysis of filtered kinetic energy transport equations. Tomographic particle image velocimetry (TPIV) and planar laser induced fluorescence (PLIF) are used to measure the 3D velocity fields and planar distribution of formaldehyde, respectively; the relevant thermodynamic properties (e.g., density and progress variable) are estimated from PLIF data. Ultimately, this work provides both an assessment of the validity of current turbulence modeling paradigms employed in aerospace relevant combustion, as well as the data necessary to develop and validate new models if required.
Verification Of Adversarially Robust Reinforcement Learning Mechanisms In Autonomous Systems

(Georgia Institute of Technology, 2022-12-07) Seo, Taehwan

Artificial Intelligence (AI) is an effective algorithm for satisfying both optimality and adaptability in autonomous control systems. However, the policy generated from the AI is black-box, and since the algorithm cannot be analyzed in advance, this motivates the performance measurement of the AI model with verification. The performance and safety of the Cyber-Physical System(CPS) are subject to cyberattacks that intend to fail the system in operation or to interrupt the system from learning by modulation of learning data. For the safety and reliability scheme, verifying the impact of attacks on the CPS with the learning system is critical. This thesis proposal focuses on proposing one verification framework of adversarially robust Reinforcement Learning (RL) policy using the software toolkit ‘VERIFAI’, providing robustness measures over adversarial attack perturbations. This allows an algorithm engineer would be equipped with an RL control model verification toolbox that may be used to evaluate the reliability of any given attack mitigation algorithm and the performance of nonlinear control algorithms over their objectives. For this specified work, we developed the attack mitigating RL on nonlinear dynamics by the interconnection of off-policy RL and on-off adversarially robust mechanisms. After that, we connected with the simulation and verification toolkit for testing both the verification framework and integrated algorithm. The simulation experiment of the whole verification process was performed with two different control problems, one is a cart-pole problem from OpenAI gym, and the other problem is the attitude control of Cessna 172 in X-plane 11. From the experiment, we analyzed how the attack-mitigating RL algorithm performed with gain varying specific adversary attacks, and evaluated the generated model performance over the changing environmental parameters.
A Methodology to Enable Concurrent Trade Space Exploration of Space Campaigns and Transportation Systems

(Georgia Institute of Technology, 2022-12-02) Prasad, Akshay

Space exploration campaigns detail the ways and means to achieve goals for our human spaceflight programs. Significant strategic, financial, and programmatic investments over long timescales are required to execute them, and therefore must be justified to decision makers. To make an informed down-selection, many alternative campaign designs are presented at the conceptual-level, as a set and sequence of individual missions to perform that meets the goals and constraints of the campaign, either technical or programmatic. Each mission is executed by in-space transportation systems, which deliver either crew or cargo payloads to various destinations. Design of each of these transportation systems is highly dependent on campaign goals and even small changes in subsystem design parameters can prompt significant changes in the overall campaign strategy. However, the current state of the art describes campaign and vehicle design processes that are generally performed independently, which limits the ability to assess these sensitive impacts. The objective of this research is to establish a methodology for space exploration campaign design that represents transportation systems as a collection of subsystems and integrates its design process to enable concurrent trade space exploration. More specifically, the goal is to identify existing campaign and vehicle design processes to use as a foundation for improvement and eventual integration. In the past two decades, researchers have adopted terrestrial logistics and supply chain optimization processes to the space campaign design problem by accounting for the challenges that accompany space travel. Fundamentally, a space campaign is formulated as a network design problem where destinations, such as orbits or surfaces of planetary bodies, are represented as nodes with the routes between them as arcs. The objective of this design problem is to optimize the flow of commodities within network using available transport systems. Given the dynamic nature and the number of commodities involved, each campaign can be modeled as a time-expanded, generalized multi-commodity network flow and solved using a mixed integer programming algorithm. To address the challenge of modeling complex concept of operations (ConOps), this formulation was extended to include paths as a set of arcs, further enabling the inclusion of vehicle stacks and payload transfers in the campaign optimization process. Further, with the focus of transportation system within this research, the typical fixed orbital nodes in the logistics network are modified to represent ranges of orbits, categorized by their characteristic energy. This enables the vehicle design process to vary each orbit in the mission as it desires to find the best one per vehicle. By extension, once integrated, arc costs of dV and dT are updated each iteration. Once campaign goals and external constraints are included, the formulated campaign design process generates alternatives at the conceptual level, where each one identifies the optimal set and sequence of missions to perform. Representing transportation systems as a collection of subsystems introduces challenges in the design of each vehicle, with a high degree of coupling between each subsystem as well as the driving mission. Additionally, sizing of each subsystem can have many inputs and outputs linked across the system, resulting in a complex, multi-disciplinary analysis, and optimization problem. By leveraging the ontology within the Dynamic Rocket Equation Tool, DYREQT, this problem can be solved rapidly by defining each system as a hierarchy of elements and subelements, the latter corresponding to external subsystem-level sizing models. DYREQT also enables the construction of individual missions as a series of events, which can be directly driven and generated by the mission set found by the campaign optimization process. This process produces sized vehicles iteratively by using the mission input, subsystem level sizing models, and the ideal rocket equation. By conducting a literature review of campaign and vehicle design processes, the different pieces of the overall methodology are identified, but not the structure. The specific iterative solver, the corresponding convergence criteria, and initialization scheme are the primary areas for experimentation of this thesis. Using NASA’s reference 3-element Human Landing System campaign, the results of these experiments show that the methodology performs best with the vehicle sizing and synthesis process initializing and a path guess that minimizes dV. Further, a converged solution is found faster using non-linear Gauss Seidel fixed point iteration over Jacobi and set of convergence criteria that covers vehicle masses and mission data. To show improvement over the state of the art, and how it enables concurrent trade studies, this methodology is used at scale in a demonstration using NASA’s Design Reference Architecture 5.0. The LH2 Nuclear Thermal Propulsion (NTP) option is traded with NH3and H2O at the vehicle-level as a way to show the impacts of alternative propellants on the vehicle sizing and campaign strategy. Martian surface stay duration is traded at the campaign-level through two options: long-stay and short-stay. The methodology was able to produce four alternative campaigns over the course of two weeks, which provided data about the launch and aggregation strategy, mission profiles, high-level figures of merit, and subsystem-level vehicle sizes for each alternative. Expectedly, with their lower specific impulses, alternative NTP propellants showed significant growth in the overall mass required to execute each campaign, subsequently represented the number of drop tanks and launches. Further, the short-stay campaign option showed a similar overall mass required compared to its long-stay counterpart, but higher overall costs even given the fewer elements required. Both trade studies supported the overall hypothesis and that integrating the campaign and vehicle design processes addresses the coupling between then and directly shows the impacts of their sensitivities on each other. As a result, the research objective was fulfilled by producing a methodology that was able to address the key gaps identified in the current state of the art.
Design of the VISORS and SWARM-EX Propulsion Systems

(Georgia Institute of Technology, 2022-12-01) Hart, Samuel T. ; Lightsey, E. Glenn

The Georgia Tech (GT) Space Systems Design Lab (SSDL) will deliver 3-D printed cold gas propulsion systems for the VISORS and SWARM-EX CubeSat formation flying missions. This report provides an overview of the working principle of these and past propulsion systems designed by the SSDL. Further information is provided about the specific designs of each of these systems and the problems encountered throughout the design process. Additionally, recommendations for improvements to future designs are outlined. An analysis of the effects of temperature on these systems is also presented.