Soft mobile robots offer distinct advantages for navigating complex terrains because of their inherent flexibility, which enables exceptional adaptability and versatility. However, their compliant bodies introduce significant challenges, such as motion uncertainties and unpredictable interactions with their environment, that are difficult to control. Furthermore, the complex dynamics of soft mobile robots complicate the realisation of full autonomy, a challenge that is further exacerbated by the limited sensing and proprioception capabilities employed in the field.

This thesis aims to advance the field of autonomous origami-enabled mobile robots, a subclass of soft mobile robots, by enhancing their capability to traverse complex terrains and improving their viability for real-world applications. Previous work from the Bio-inspired Adaptive Morphology Laboratory (BAM Lab) developed an Origami-Enabled Soft Crawling Autonomous Robot (OSCAR) in pursuit of this goal. OSCAR is a novel soft mobile robot that leverages origami-inspired mechanisms to mimic the crawling motion of caterpillars. Hence, building upon that foundation, this work enhances OSCAR’s capabilities by enabling traversal of complex three-dimensional spaces without relying on external sensors, paving the way for implementation of truly autonomous navigation.

OSCAR’s mechanical stability is first enhanced with a double-celled design, and vertical actuation is introduced by employing four origami towers per cell. These upgrades improve stability, maneuverability, and locomotion range, enabling complex three-dimensional terrain traversal in the updated version called the Slinky Origami-Enabled Soft Crawling Autonomous Robot (SOSCAR). Afterwards, control systems are developed to realise the new mechanical design and demonstrate vertical obstacle avoidance. Finally, internal sensing mechanisms, using Time-of-Flight distance sensors and Inertial Measurement Units, are integrated to provide proprioception, or self-awareness, that enable closed-loop positional feedback control. Whereas previous versions of OSCAR relied on external sensors for control, all sensing in SOSCAR is fully integrated onboard the robot.

Ultimately, this thesis presents a soft mobile robot that integrates the necessary elements for future implementation of autonomous navigation in complex three-dimensional terrains. As a result, it advances the real-world readiness of origami-enabled robots and highlights their potential for operating in challenging environments.

The International Moth is a small racing sailboat that can reach top speeds of 35 knots (18 m/s), due to its use of hydrofoils, which lift the entire hull clear of the free surface. The hydrofoils replace the hull as the primary generators of hydrodynamic forces within the vessel system, and in turn, heavily drive the overall performance of the vessel. Optimizing the shape and planform of the foils is a key to achieving race-winning designs. However, hydrofoiling sailboats are highly coupled systems that operate in two simultaneous fluid media, and a change in foil configuration can have cascading effects on the overall vessel state. Thus, a design framework is formulated that allows foil designs to be evaluated within a 6 degree of freedom velocity prediction program (VPP). The framework integrates gradient-based shape optimization tools in 2 and 3 dimensions. Evaluation of the framework demonstrates functionality for design optimization independent of the VPP, but the presented approach to modeling hydrodynamic forces within the VPP requires improvement in order to produce meaningful results that can inform design decisions.

Decarbonizing Brazil is crucial for reducing global greenhouse gas emissions. To that end, the Net-Zero Brazil (NZB) modeling study aims to provide viable pathways for the country to achieve net-zero emissions by 2050. The modeling will be done with unprecedented spatial, technological, and temporal resolution. It relies on a least-cost, multi-sector optimization model being developed by the Princeton ZERO Lab called MACRO. Given Brazil’s prominence in biofuels and land-use challenges, a strategic approach to bioenergy deployment is essential. My study presents a high-resolution bioenergy supply chain optimization model, Downscale, designed to integrate into NZB to determine cost-effective bioenergy production, processing, and distribution pathways at fine spatial, temporal, and technological resolutions. A key feature is its downscaling capability, which enhances MACRO by translating state-level energy system results into actionable strategies for local deployment. Downscale is a mixed-integer linear programming model that optimally locates bioenergy crops, conversion facilities, and transportation while incorporating economic, environmental, and land-use constraints within municipalities. It is a myopic optimization model with no look-ahead, called at every time step of MACRO optimization. Cost-supply data for biomass resources and techno-economic characteristics of a portfolio of conversion technologies were gathered at the municipality-level (5570 municipalities in Brazil) for future use in MACRO via state-level aggregation. Downscale was then tested for Mato Grosso do Sul, a key biofuel-producing state. Four scenarios were analyzed: uniform demand growth for bio-derived energy carriers, low environmental protection (allowing bioenergy crop production in the Pantanal region), modest electrification of energy demands, and high electrification of demands. Results indicate that strategic infrastructure expansion can meet rising bioenergy demands while minimizing costs and environmental impacts, while also highlighting trade-offs in land-use decisions and resource allocation. This model provides actionable insights for policymakers and investors while serving as both an enhancement to MACRO and NZB and a standalone tool for downscaling optimization problems.

This project seeks to design one possible solution to the issue of accessibility on theatrical stages, as well as the issue of limited design configurations. This report goes through multiple iterations to discover the best materials from which to construct a functional automated platform for the stage. An emphasis is placed on ensuring that the platforms can tilt to form an ADA-accessible ramp from the ground to the stage. Safety factors are also emphasized, considering the machine must lift human payloads and therefore must be designed to these standards. The suggested technology, a linearly actuated Stewart Platform, is not currently used in the theater world, and this concept could open up many new possibilities for designers as well as accessibility.

Technology is ever-evolving, and its usefulness in sports training applications has increased with each evolution. Coaches and teams turn to technology to analyze player movements, break down game footage, and improve athletic performance. The sport of ice hockey has seen rapid player development over its history, with the game becoming faster, player equipment constantly improving, and the sport more popular and accessible than ever [4]. However, despite the abundance of training technology available for ice hockey players, one position still lags behind in development: the goalie. Ice hockey goalies often develop at a slower pace than their skater teammates due to a lack of expert coaching and resources at the lower levels, as well as the specialized equipment required, which makes the position more expensive and creates barriers to entry [19]. Additionally, goalies see the most growth in their playing ability when they get meaningful ice and game time. This thesis explores various designs of automated ball pitching and launching machines, as well as previous attempts to develop hockey puck shooting machines for goalie training. It also investigates why these earlier designs failed to gain widespread adoption within the goalie training community. Drawing on these insights and market research conducted for this project, the goal of this senior thesis is to design and build a functional puck launcher that is portable, versatile in capability, and can operate both on and off the ice, while being easily transported between different training locations.

Studying aquatic organisms requires close-range observation that minimizes disruption to natural behaviors. Bio-inspired robotic fish offer a promising low-impact platform for underwater exploration, but achieving depth control in such systems remains challenging. This thesis presents the design, development, and evaluation of a depth control mechanism for the BlueKoi, a bio-inspired robotic fish developed by the SSR Nagpal Lab at Princeton University. Previously lacking vertical maneuverability, the BlueKoi achieved forward swimming and turning through a tuna-inspired tail actuation system. After evaluating several depth control strategies, including a sliding weight, a buoyancy control unit, and upward-facing propulsion, actuated pectoral dive planes were selected given the BlueKoi’s existing size and mass constraints. The final design features three main components: a magnetic coupling “shaft” that transmits actuation from inside the body to an external dive plane through the intact body wall, achieving complete waterproofing; a modular internal bevel gear-servo assembly achieving a dive plane rotation range of -45 to +35°; and an external hydrofoil mounting interface enabling easy hydrofoil interchangeability. Eight hydrofoil variants based on NACA 0012, 0015, 0018, and 0021 profiles of varying span and chord lengths were evaluated in a water channel test at the BlueKoi's approximate minimum swimming speed (0.19 m/s). The system demonstrated reliable waterproofing across sixteen trials, and force measurements confirmed that negative angles of attack consistently produced negative lift, establishing the dive planes' capacity to initiate dives. The NACA 0018 profile with increased span performed best. A free-swim test further confirmed effective depth control, with the BlueKoi executing repeatable dives and ascents at multiple speeds. Depth changes appeared driven primarily by sustained lift forces rather than body pitch, indicating that continuous actuation is required. These results establish the feasibility of actuated dive planes as a depth control mechanism for the BlueKoi, and motivate future work on pitch characterization, actuation refinement, and complementary approaches such as dynamic buoyancy control.

Bridges are highly susceptible to aerodynamic and hydraulic loading due to their proximity to water bodies and exposure to extreme weather events. With climate change driving an increase in the frequency and intensity of such events, understanding the influence of structural porosity on the aerodynamic behavior of truss bridges has become increasingly important. This study experimentally investigates the aerodynamic drag characteristics of truss bridges in a wind tunnel, with the aim of characterizing the relationship between drag coefficient Cd, Reynolds number Re, and structural porosity 𝛽 in the subcritical flow regime (Re = 10^4 - 10^5). Four bridge configurations were developed based on the Picassent truss bridge, which was damaged during the 2024 Valencia flood in Spain. These configurations represent common truss topologies: Pratt (𝛽 = 75.85%), Warren without Verticals (𝛽 = 75.75%), Warren with Verticals (𝛽 = 69.99%), and a partially covered Warren simulating debris accumulation (𝛽 = 34.98%). An inverse relationship between drag coefficient and porosity was observed, with drag decreasing as porosity increases, consistent with existing literature. Additionally, the effective Reynolds number, which incorporates structural porosity, was found to provide a more physically meaningful scaling parameter for comparing porous structures of varying porosity. These results highlight the importance of incorporating structural porosity into the aerodynamic design of large truss bridge structures. As this study was conducted at low wind speeds (6-14 m/s), it serves as a preliminary investigation, with the bridge models acting as prototypes for future testing in the High Reynolds Number Test Facility (HRTF) to explore behavior in the supercritical flow regime.

For decades, aircraft have leveraged ground effect—a phenomenon that increases lift and decreases drag on surfaces near the ground—to enhance flight efficiency. However, its influence on low-Reynolds-number flight, relevant for small unmanned aerial vehicles (sUAVs), is vastly underexplored. Notably, small organisms that frequently hunt, cruise, and forage near the ground appear to illustrate a rich source of inspiration for driving better sUAV design in this regime.

This paper thus investigates the performance of a bat-inspired wing, “Chiropter”, modeled after the well-studied species Cynopterus brachyotis, when subjected to an engineered ground effect boundary.

A comparative study between a conventional NACA 2414 airfoil and bat-inspired planforms was conducted in Princeton University’s wind tunnel. Force measurements were acquired in the wind tunnel’s test section (1.2 x 1.2 m cross section, 1 m long) in the low Reynolds number regime while the wings were subjected to an engineered ground effect boundary. From testing three bat bioinspired planforms, including a rigid wing, a flexible wing, and an actuated flexible wing for downstroke-upstroke cycles, it was determined that flexible wings generally trade efficiency and stability for enhanced lift. Through the trials, ground effect was observed to raise the lift and alter the drag progression for the wings, such that more flexible wings may observe greater lift at small angles of attack and large moments at large angles of attack. Furthermore, for unsteady flapping flight, ground effect vastly alters the surrounding airflow, restricting downwash development and the overall deflection of the wing, while still roughly preserving aerodynamic efficiency during the powered downstroke.

The results of this study not only improve the understanding of how bats leverage ground effect to alter their performance, but they also may be used to inform how engineers can design more efficient small-scale aerial vehicles for a variety of use cases, including environmental monitoring, search & rescue, and agricultural applications.

Difference Engine No. 2, designed by Charles Babbage between 1847 and 1849, is a mechanical calculator capable of tabulating seventh-order polynomials to thirty-one decimal places. Construction of the Engine, which was never built during Babbage's lifetime, was fully completed by the London Science Museum in 2002, proving the viability of his intricate design based on the method of finite differences. This project presents an open-access, 3D-printed kit of Difference Engine No. 2's Calculation Section, scaled to a computational limit of third-order polynomials to four digits. The kit measures roughly two feet long, nine inches tall, and nine and a half inches wide, sitting at a scale sufficiently small for a hobbyist 3D-printer but sufficiently large to visualize the complete mechanisms of the Calculation Section. Challenges included a lack of dimensions and manufacturing detail in the original drawings, and adaptation to the tolerance and friction limitations of 3D-printing. Assembly of the final kit requires no tools beyond super glue and pliers, and all components are 3D-printed with the exception of the springs from twelve Pilot G-2 Gel pens. The CAD and 3D-print files, as well as assembly documentation, have been made publicly available on MakerWorld, providing an entry point for educators and enthusiasts alike to gain hands-on experience with early mechanical computing.

This thesis focuses on the design for an autopilot able to aid in landing during microbursts, weather events that can complicate landing approaches due to large wind shears. If an aircraft traverses through the center of a microburst, it experiences an almost instantaneous change from a headwind to a tailwind that can result in total wind shear exceeding 100 kn. Aircraft on approach to land face particular issues with this event due to the microbursts' increased strength near the ground, combined with the aircraft's low speeds and altitude, causing several well-documented airline crashes. A design for an autopilot of a Cessna 172 trainer aircraft is presented that enables the aircraft to safely land despite encountering a microburst near the airport. This autopilot allows the aircraft to be pushed by the high winds away from the microbursts' center, maneuvering around the strongest winds before returning to alignment with the runway's approach. The aircraft is modeled and its aerodynamic data collected to allow for a high-fidelity simulation of this movement to serve as a pedagogical tool in microburst evasion for student pilots. The open-source flight simulator FlightGear is mapped to the simulation for viewing and validation of suggested evasion methods. The autopilot suggests not fighting heading, allowing the wind to push the aircraft laterally. The success of these suggestions were verified in manual control tests in the simulated environment.

In the field of small-scale robotics, motion typically involves a tradeoff between strength and speed. However, many small animals are still capable of achieving extremely powerful motions using latch-mediated spring actuation (LaMSA). Such processes are often internal and difficult to study, but there is a widely-available exception: click beetles. These cosmopolitan beetles can jump without using legs or wings, through the actuation of a mechanism in the hinge between body segments.

The objective of this work is to develop a biologically relevant robotic model organism (RMO) that characterizes this actuation mechanism, integrating morphologically accurate body segments with a LaMSA hinge developed by prior research at the Bio-inspired Adaptive Morphology Lab at Princeton University. This design enables controlled investigation of the mechanical principles governing loading and unlatching for click beetle LaMSA. Potential energy is stored through a combination of linear and torsional springs, while a geometric latch mediates release. The final system is actuated using an SMA actuator.

Overall, this work presents an experimental framework for studying LaMSA systems through a robotic model. By bridging the gap between biology and engineering, the designed RMO enables isolated exploration of parameter spaces beyong those found in live click beetles. The insights gained from this study contribute to the design of high-power, small-scale robotic systems and advance understanding of the mechanical principles underlying LaMSA in both biological and engineered contexts.

This thesis explores the design and simulation of a control system that would enable omnicopter drones to simulate close proximity spacecraft rendezvous maneuvers. Close proximity rendezvous simulations typically exist either as simplified online models that lack real-world dynamics or as prohibitively expensive physical testbeds. By leveraging the six degrees of freedom (6-DOF) capability of omnicopters—drones with non-planar rotor configurations allowing complete spatial movement developed by Dario Brescianini and Raffaello D’Andrea—this research provides a cost-effective middle ground between purely digital and full-scale physical simulations. The study implements relative orbital motion using the Clohessy-Wiltshire equations to model the dynamics between a ”chief” and ”deputy” omnicopter, simulating target and approaching spacecraft respectively in close proximity operations. A cascaded control architecture that separately handles attitude and position control enables the simulated deputy omnicopter to approach and maintain specific poses relative to the chief within small distances, mirroring the final critical phase of spacecraft rendezvous. Using MATLAB, the research validates this control strategy through simulation, demonstrating its effectiveness for precise close proximity maneuvers. This work’s contributions lie in: (1) implementing close proximity orbital rendezvous control for omnicopters in simulation, (2) providing improved documentation of omnicopter capabilities to address gaps in publicly available resources, and (3) establishing a foundation for future physical implementation and testing. While physical deployment remains outside the scope of this thesis, the comprehensive modeling and simulation work presented here creates a viable pathway between theoretical spacecraft dynamics and accessible hardware implementation for future research in close proximity operations.

The rapid growth of the commercial space sector, especially CubeSats, has made lowcost, compact sensor hardware an increasingly practical option for spacecraft applications, yet the gap between terrestrial development and space qualification remains a critical challenge. This thesis addresses that gap through the design, assembly, and testing of two experimental payloads developed with the MaxIQ Space program, in support of the Princeton TigerSats Laboratory’s development of a low-cost 1U CubeSat.

The first payload targets a stratospheric flight aboard a NASA High-Altitude Balloon, validating commercial-off-the-shelf infrared horizon sensors to detect Earth’s horizon and compute spacecraft attitude estimates under near-space conditions. The second payload targets orbital deployment aboard the International Space Station, collecting in-situ environmental measurements to characterize the payload’s local operational conditions and validate the viability of commercial sensors in a true microgravity environment as well as their launch survivability.

Across both experiments, design prioritized minimal volume, mass, and power consumption to adhere to the design requirements and constraints. Candidate sensors were evaluated through ground testing and iterative PCB design before delivery for launch, aiming to advance their Technology Readiness Level for future integration into the TigerSats CubeSat.

This project presents the Compliant Robotic Extruded Hand (CRxH), a 3D-printed monolithic compliant robotic hand designed as a low-cost and easily replicable platform for prosthetic research and expendable manipulation in hazardous environments. The primary goal is to achieve human-like finger flexion through compliant underactuated structures that bend under low force while preserving structural integrity and elastic recovery (“shape memory”) over repeated use. The hand was iteratively designed to localize bending in compliant regions akin to the human hand and to prevent permanent deformation after repeated actuation. The anatomically inspired tendon routing facilitates realistic multi-joint finger curling, enables coordinated hand motion, and decreased assembly complexity. Strength tests found the hand to have a static strength capacity of at least 75 lbs, with each finger supporting at least 25 lbs. Qualitative dexterity tests found CRxH capable of mirroring numerous human actuation patterns and grasping various objects. The final hand prototype achieves 14 degrees of freedom, approaching the 21 degrees of freedom of the human hand. This repeatable motion is achieved without permanent plastic deformation. Thermoplastic Polyurethane (TPU) was chosen for the hand to support flexibility and compliance, while Polylactic Acid (PLA) was used for the rigid forearm motor housing. Both filaments can be 3D-printed with minimal client-side debugging. CRxH serves as an important stepping stone for future work on compliant, underactuated mechanisms for robotic manipulators by opening the door to improvements in accessible and rapid monolithic 3D-printed designs, highlighting grasping capability, closed-loop control, and autonomy. CRxH is able to provide comparable dexterity and a higher strength performance at a fractional cost compared to state-of-the-art commercial and research designs, accomplished in a single print.

This project presents the Compliant Robotic Extruded Hand (CRxH), a 3D-printed monolithic compliant robotic hand designed as a low-cost and easily replicable platform for prosthetic research and expendable manipulation in hazardous environments. The primary goal is to achieve human-like finger flexion through compliant underactuated structures that bend under low force while preserving structural integrity and elastic recovery (“shape memory”) over repeated use. The hand was iteratively designed to localize bending in compliant regions akin to the human hand and to prevent permanent deformation after repeated actuation. The anatomically inspired tendon routing facilitates realistic multi-joint finger curling, enables coordinated hand motion, and decreased assembly complexity. Strength tests found the hand to have a static strength capacity of at least 75 lbs, with each finger supporting at least 25 lbs. Qualitative dexterity tests found CRxH capable of mirroring numerous human actuation patterns and grasping various objects. The final hand prototype achieves 14 degrees of freedom, approaching the 21 degrees of freedom of the human hand. This repeatable motion is achieved without permanent plastic deformation. Thermoplastic Polyurethane (TPU) was chosen for the hand to support flexibility and compliance, while Polylactic Acid (PLA) was used for the rigid forearm motor housing. Both filaments can be 3D-printed with minimal client-side debugging. CRxH serves as an important stepping stone for future work on compliant, underactuated mechanisms for robotic manipulators by opening the door to improvements in accessible and rapid monolithic 3D-printed designs, highlighting grasping capability, closed-loop control, and autonomy. CRxH is able to provide comparable dexterity and a higher strength performance at a fractional cost compared to state-of-the-art commercial and research designs, accomplished in a single print.

This project presents the Compliant Robotic Extruded Hand (CRxH), a 3D-printed monolithic compliant robotic hand designed as a low-cost and easily replicable platform for prosthetic research and expendable manipulation in hazardous environments. The primary goal is to achieve human-like finger flexion through compliant underactuated structures that bend under low force while preserving structural integrity and elastic recovery (“shape memory”) over repeated use. The hand was iteratively designed to localize bending in compliant regions akin to the human hand and to prevent permanent deformation after repeated actuation. The anatomically inspired tendon routing facilitates realistic multi-joint finger curling, enables coordinated hand motion, and decreased assembly complexity. Strength tests found the hand to have a static strength capacity of at least 75 lbs, with each finger supporting at least 25 lbs. Qualitative dexterity tests found CRxH capable of mirroring numerous human actuation patterns and grasping various objects. The final hand prototype achieves 14 degrees of freedom, approaching the 21 degrees of freedom of the human hand. This repeatable motion is achieved without permanent plastic deformation. Thermoplastic Polyurethane (TPU) was chosen for the hand to support flexibility and compliance, while Polylactic Acid (PLA) was used for the rigid forearm motor housing. Both filaments can be 3D-printed with minimal client-side debugging. CRxH serves as an important stepping stone for future work on compliant, underactuated mechanisms for robotic manipulators by opening the door to improvements in accessible and rapid monolithic 3D-printed designs, highlighting grasping capability, closed-loop control, and autonomy. CRxH is able to provide comparable dexterity and a higher strength performance at a fractional cost compared to state-of-the-art commercial and research designs, accomplished in a single print.

Understanding the transport fate of small particles injected from a surface into a crossflow is relevant to applications ranging from debris particulate management in fusion reactors to aerosol dispersion in urban settings. The key question is whether an injected particle remains confined near the wall or penetrates deep into the flow, and how this outcome depends on the particle’s inertia and injection speed. This study develops a computational algorithm that maps transport outcomes across a two-dimensional parameter space defined by the Stokes number St and the normalized injection velocity v0/Uref, producing regime maps that identify the boundaries between near-wall confinement, partial penetration, and midplane crossing. The algorithm was first validated against an exact analytical solution in laminar Poiseuille flow, achieving agreement to within 0.14%. The laminar regime boundaries followed power laws with exponents of −0.81 and −0.85, departing from the analytically predicted exponent of −1 due to coupling between the streamwise and wall-normal particle motion introduced by the velocity profile. The algorithm was then applied to a turbulent carrier flow using eleven frozen snapshots of a DNS channel flow at Reτ = 5200 from the Johns Hopkins Turbulence Database. Within the frozen-field ensemble used here, the turbulent ensemble boundaries have exponents of approximately −0.84 and lower amplitudes than their laminar counterparts, indicating that on average turbulent velocity fluctuations slightly enhance particle penetration beyond what the laminar baseline predicts. The midplane crossing probability transitions sharply near St = 0.179, within the turbophoresis-dominant range. A non-dimensional analysis of secondary forces shows that gravity shifts the regime boundaries by a predictable amount linear in St, confirmed numerically to within 1.13% for St ≤ 0.10, while the Lorentz force on charged particles is negligible for micron-scale grains at typical magnetic field strengths.

Offshore wind farms have many advantages over traditional energy systems and great potential for power generation. However, researchers are currently facing challenges in developing technologies for offshore wind farm analysis that are both computationally efficient and highly accurate. This study aims to develop a surrogate model for analyzing farms that is more efficient than existing computational models but maintains higher accuracy. It also aims to subsequently use this model to design offshore wind farms optimized for both power generation and cost of energy. Computational Fluid Dynamics (CFD) simulations were run for a design space with six features and one output and used to train a statistically robust ensemble model. This model was trained to determine relationships between the parameters and power output, tuned to achieve the greatest accuracy, and tested to ensure the legitimacy of predictions. Finally, using buoy data and Bayesian optimization, this model was used to design farms for various regions through the determination of a set of optimal structural parameters. A model with training and testing R2 values of 0.98 and 0.99 was successfully developed. It was found that, consistently among the four different locations tested with varying environmental conditions, there was an overarching optimal design: staggered orientation, 9-diameter spacing, 115 m-high hubs, and 175 m-rotor diameters. These structural parameters produced roughly 5.52−5.65 MW/m4 for each location. This consistency greatly simplifies wind-farm design going forward.

This thesis presents a modular, fuel-optimal framework for autonomous reconfiguration of satellite swarms in low Earth orbit. Built using convex optimization and Clohessy-Wiltshire dynamics, the system enables agents to maneuver into desired formations while minimizing total DeltaV. It supports both centralized and event-triggered control modes, and includes logic for fault-aware role reassignment when agents fail or drift off-nominal. The architecture is designed for extensibility and validated under orbital parameters from NASA’s Starling mission, anchoring the simulations in a realistic mission context.

Beyond idealized dynamics, I extended the framework into a nonlinear regime, incorporating repulsion-based collision avoidance and full orbital propagation. Although early implementations using MATLAB’s fmincon solver failed to resolve hard-constrained formulations, a successful reconfiguration was later achieved through soft-penalized collision avoidance. This final nonlinear simulation demonstrated precise formation change under actuator and safety constraints, revealing tradeoffs between feasibility, fuel cost, and control fairness in high-dimensional swarm settings.

Across eight original simulations, I validated control strategies that are adaptive, resilient, and fuel-efficient—ranging from passive drift modeling to fault-tolerant reconfiguration, perturbed execution, and constrained nonlinear optimization. These simulations, along with the full source code and CVX routines, are publicly released on GitHub at github.com/sabrinanicacio/satellite-swarm-thesis. This thesis delivers one of the first open-source testbeds for mission-relevant satellite swarm reconfiguration using both CW-based convex planning and exploratory nonlinear control.

Together, these contributions provide a practical foundation for future work in large-scale, autonomous satellite maneuvering. By revealing the architecture-level tradeoffs between fuel use, feasibility, and safety enforcement, this project bridges a critical gap between theoretical swarm control and operational flight software.

Ammonia is one of the most promising carbon-free energy carriers, but its slow reaction kinetics, long ignition delay time, and narrow flammability range make direct detonation initiation difficult and limit its use in pressure-gain combustion systems. Reactive co-fuels offer a possible way to overcome these challenges. However, the deflagration-to-detonation transition (DDT) behavior of ammonia-dimethyl ether (NH₃/DME) mixtures has not yet been experimentally investigated, despite DME’s well-established role as a strong low-temperature ignition promoter. This thesis addresses that gap through a combined computational and experimental study of NH₃/DME/O₂/Ar mixtures in a microchannel. NASA CEA and Cantera are first used together to evaluate different fuel compositions based on detonability, ignitability, and equilibrium emissions, narrowing a broad design space into a practical experimental matrix. To support reliable diagnostics across multiple observation windows, an automated traversal system was developed to maintain optical alignment during imaging. Flame propagation and wave-structure evolution were then captured using synchronized high-speed imaging and shadowgraphy. Results at ϕ = 1.0 show repeatable DDT in the 90/10 and 80/20 DME/NH₃ blends, while the 75/25 composition marks the practical ignition limit of the current experimental configuration. Compositions exceeding approximately 25% ammonia failed to ignite under the tested conditions. Across all ignitable blends, increasing ammonia content progressively delays both first-stage ignition and DDT onset, while also lowering peak flame-tip velocity. Despite this reduction in reactivity, successful transition is still consistently achieved in the ignitable cases. This demonstrates that ammonia can reliably undergo the full DDT process when mixed with DME. These results show that DME acts as an effective enabling co-fuel, offsetting ammonia’s reactivity limitations and bringing meaningful NH₃ fractions into the detonable regime. Extending this work to fuel-rich NH₃/DME mixtures, where computational analysis predicts higher detonation velocities and potentially earlier DDT onset, is a promising next step for accessing stronger detonation regimes and even higher NH₃ fractions. Overall , this thesis establishes NH₃/DME blends as a promising mixture for detonation-based combustion and supports ammonia’s role as a practical carbon-free fuel for pressure-gain propulsion and power systems.