Masters Theses
Permanent URI for this collection
Browse
Recent Submissions
- UAV-Based Deep Learning for Multi-Crop Plant Counting, Geolocation, and Field-Scale Decision SupportSeetharaman, Ragul (Virginia Tech, 2026-07-27)Accurate plant stand assessment is essential for early-season crop management because plant population, emergence uniformity, and spatial distribution influence crop growth, yield potential, and management decisions such as replanting. Traditional plant counting is commonly performed through manual field assessment, but manual counting is time-consuming, labor-intensive, and difficult to repeat consistently across large fields, multiple cropping systems, and different imaging conditions. Recent advances in unmanned aerial vehicle imagery and deep learning have created new opportunities for automated plant detection and counting. However, many existing workflows remain crop-specific, model-specific, or dependent on programming-based processing steps that limit practical field use. This thesis presents a manuscript-based study on UAV-based deep learning for automated plant counting, geolocation, and field-scale visualization across multiple crops. The first manuscript evaluates a standardized object detection pipeline for plant counting across faba bean, strawberry, cotton, and tobacco. The pipeline applies image preprocessing, tiled inference, and consistent evaluation procedures to compare YOLO-style and DETR-style object detectors under varying crop morphology, plant density, background texture, and illumination conditions. The study benchmarks YOLOv12, YOLO26, RT-DETR, and RF-DETR using detection-level metrics and count-level agreement metrics. Results show that YOLOv12 achieved the strongest detection performance across the crop datasets, while count reliability varied by crop due to factors such as canopy overlap, shadows, dense rows, small targets, and background texture. The second manuscript extends the detection workflow into a graphical user interface-based plant-counting system. The developed PlantCountGUI integrates trained object detection models with orthomosaic tiling, tile-level inference, coordinate transformation, duplicate detection removal, count summarization, visualization, and GIS-compatible export. Users can upload georeferenced UAV orthomosaics, select processing parameters, run detection, inspect plant distribution maps and density heatmaps, and export CSV and shapefile outputs. By converting image-based detections into projected field coordinates and presenting results through an accessible interface, the system reduces the gap between model development and practical deployment for agricultural researchers, extension personnel, and growers. Together, these manuscripts contribute a standardized multi-crop evaluation framework and an applied software tool for UAV-based plant stand assessment. The thesis demonstrates that deep learning-based plant counting can support precision agriculture when detection accuracy, count reliability, geospatial processing, and usability are considered together.
- The Art of Diminishing: Poetic Geometry and Generative Proportion in Tiered Temple of Kathmandu ValleyAyer, Yusna (Virginia Tech, 2026-11-10)Architecture is often understood through its cultural, historical, and symbolic meanings, yet the physical forms through which these meanings are expressed are governed by systems of geometry, proportion, and order. In the architecture of the Kathmandu Valley, the tiered temple represents a distinctive synthesis of cultural tradition and formal organization. While these temples have been studied through their history, symbolism, religious significance, structural behavior and spatial organization, the geometric relationships that generate and organize their vertical form remain less explicitly understood. The complex appearance of tiered temples suggests an underlying order in the relationship between the plinth, walls, cella, and successive roof tiers. These relationships are not necessarily defined by a single fixed dimension or universally documented rule, but may instead emerge through proportional relationships, recurring geometric conditions, and transformations of form. Understanding these relationships raises the question: can the geometry of tiered temples reveal a systematic logic through which their architectural form was established and transformed? This research investigates the geometry and proportional relationships of selected tiered temples of the Kathmandu Valley through observation, photographic measurement, digital reconstruction, and parametric modeling. By examining the relationship between width, height, tier progression, roof slopes, datum conditions, and the proportions of the temple's major components, the research seeks to identify recurring geometric relationships within the vertical composition of these structures. A Grasshopper-based parametric model is further used to reconstruct and manipulate these relationships, allowing the observed geometric conditions to be explored through controlled transformations of the temple form. Rather than reducing traditional architecture to a fixed mathematical formula, this research approaches geometry as a means of understanding how architectural form may carry, preserve, and transform knowledge across time. The study seeks to re-discover the underlying order within the visual complexity of Kathmandu Valley's tiered temples and to demonstrate how the study of geometry can provide a bridge between cultural tradition, architectural form, and contemporary methods of digital exploration.
- Scam Shield: A Hierarchical Framework for Robust and Efficient Adversarial Scam DetectionChang, Chen-Wei (Virginia Tech, 2026-11-05)Can we trust large language models (LLMs) to accurately detect scams? This thesis investigates the vulnerabilities of LLMs when facing adversarial scam messages. We created a dataset with fine-grained labels, including original scam messages, adversarial scam messages, and non-scam messages. These labels allow us to examine detection performance across different scam types in addition to the binary classification task. We evaluated GPT, Claude, and LLaMA on this dataset. Under the selected rewriting conditions, the evaluated models miss scam messages even when the deceptive intent is preserved. To address this problem, we propose a Hierarchical Scam Detection System (HSDS). The system combines four lightweight classifiers, Decision Tree, XGBoost, Random Forest, and K-Nearest Neighbors, with a fine-tuned LLaMA 3.1 8B Instruct model. The classifiers first predict whether each message is a scam. When all four agree, the system returns their prediction. When they disagree, the message is sent to the language model. We fine-tuned this model with Low-Rank Adaptation (LoRA) using a domain-specific adversarial dataset. The fine-tuned model achieved higher detection scores with zero-shot prompting than the evaluated few-shot baselines. The full HSDS achieved an accuracy of 0.90 and an F1-score of 0.90, with higher scores than the evaluated proprietary LLM baselines. It also reduced average inference time by approximately 56.7% compared with the fine-tuned model alone. Romance scams remained difficult to detect. Further work could examine whether semantic and contextual features improve detection of these emotionally manipulative messages.
- Physics-Informed Machine Learning for Predicting Rubber–Surface Friction from Dynamic Friction Tester Measurements: Load–Speed Interaction, Load, Roughness, and Material ResponseHamedi, Sajad (Virginia Tech, 2026-09-11)The interaction between tires and the road is one of the main factors determining how safely and effectively a vehicle can operate. Predicting this friction, however, is far from straightforward because it depends on several variables, such as structural and material properties of tire, road roughness, tire slip, applied load, temperature, and velocity, that interact in complex and nonlinear ways. Existing models can handle simplified conditions but often fall short when applied to real driving situations. This thesis addresses the problem through physics-informed machine learning. Approximately 600 steady-state friction observations produced by a Dynamic Friction Tester (DFT), collected by Su [8] at the CenTiRe Laboratory at Virginia Tech, were assembled and used to evaluate a range of machine learning techniques. Eight rubber compounds were tested across three standardized abrasive surfaces spanning coarse to fine texture. Random Forest, Support Vector Regression, Artificial Neural Network, and a Hybrid Stacking Ensemble were trained and compared against the measured friction coefficients. Because the measurements were made on abrasive papers rather than on pavement, the friction coefficients span a wider and higher range than a tire encounters in service; the models are therefore calibrated to a DFT-on-abrasive regime rather than to a road regime. The central contribution of this work lies in how the learning problem is posed rather than in the volume of data analyzed. Raw experimental inputs such as sliding velocity and viscoelastic loss modulus are coupled through the excitation frequency relationship 𝜔 ≈ 𝑞𝑉s described in Persson’s viscoelastic friction theory [2,59–61]. The feature basis was therefore re-expressed through two mechanism-guided interaction terms: Fn×Vs, a power-dimensional load–speed interaction, and Ts×E″, a strength–damping coupling formed from the compound’s maximum tensile strength and viscoelastic loss modulus. The physical interpretation of the first term is deliberately limited: Fn×Vs is not the actual frictional power delivered at the interface, which is 𝐹f𝑉s= 𝜇𝐹n𝑉s. Su [8] likewise expresses sliding work using friction force, 𝑊 = 𝐹f𝑠. The second term is also an engineered interaction rather than a derived physical quantity; its units, Pa², are not those of energy or dissipation. Models trained on this reparameterized feature set achieved high accuracy while producing feature-importance rankings that are interpretable in tribological terms. Overall, the study shows that a feature basis built from tribological reasoning yields accurate friction predictions whose importance structure is interpretable in mechanistic terms. A matched raw-feature control was subsequently trained on {Dp, Ts, Fn, Vs, E″, grit} under the same split, preprocessing and hyperparameters, and all four architectures were re-evaluated under leave-one-compound-out validation. The control shows that the composite terms are not universally more accurate: across twenty repeated splits they give the Random Forest a small but reproducible gain (ΔR² = +0.005, p = 0.024), while the raw basis is better for SVR and ANN and indistinguishable for the Hybrid Ensemble. Compound-holdout validation shows that the Random Forest predicts entirely unseen formulations with pooled R² = 0.921, including R² = 0.858 when trained on compounds A–G and tested on the withheld compound H. The claim advanced here is therefore one of interpretability, physical consistency and demonstrated material generalization rather than of universal predictive superiority. The framework is a step toward friction models that can be audited against theory as well as against data.
- Exploring Organogenesis in American ChestnutKocz, Sebastian James (Virginia Tech, 2026-10-29)American chestnut (Castanea dentata) was once a major presence in the forests of the Eastern United States. However, fifty years after chestnut blight, a canker disease caused by Cryphonectria parasitica, was first observed in the Bronx Zoo in 1904, the tree was considered functionally extinct. Efforts to increase disease resistance in the species via genetic engineering are ongoing, however they would be aided by improvements in the efficiency and speed of in vitro regeneration. To begin optimizing a de novo organogenesis regeneration protocol for C. dentata, we tested seven different callus induction media formulations as well as Agrobacterium-mediated transformation of explants with ten different morphogenic regulators (MRs) that had been reported to increase shoot induction when overexpressed in other plant species. We identified a high-performing callus induction media (WPM with 2 mg BAP and 0.5 mg NAA) that produced callus on 87.5% of C. dentata leaf explants after being kept in a dark environment for 3 weeks. In our three transformation experiments with MRs, while we found some MRs to cause significantly more callus to form than others, we found no significant improvement in callus induction in explants transformed with MRs compared to GFP-transformed controls. In some cases, the GFP-transformed control performed significantly better. Across both studies, four adventitious shoots were observed. These represent the first reported shoots formed by C. dentata through de novo organogenesis and give us optimism that with more optimization, a consistent protocol can be generated.
- Stabilized vs. Unstabilized Barrier Islands: How Shoreline Engineering Shapes the Habitat, Prey, and Distribution of Red Knots, Black‑bellied Plovers, and Piping Plovers after Five YearsAlvino, Camille Rose (Virginia Tech, 2026-10-23)Climate change-driven storms and sea-level rise are altering Mid-Atlantic coastlines at accelerating rates, particularly in barrier-island landscapes. Migrating shorelines threaten coastal infrastructure and create pressure on imperiled shorebirds that depend on high-quality foraging grounds year-round. Shoreline-stabilizing techniques, including beach nourishment, breakwaters, and revetments, are used to mitigate coastal erosion and protect local infrastructure, but their direct and indirect impacts on shorebird populations and their prey are poorly understood. The Virginia barrier island system, recognized for its ecological importance to shorebirds, is currently undeveloped except for one island, Wallops Island, which has undergone various stabilization measures to protect critical infrastructure. This system provides a unique opportunity to expand existing knowledge of how ecological and geomorphic factors influence shorebird foraging habitat use in the context of anthropogenically altered shorelines. The goals of my thesis were to quantify the ecological and geomorphological drivers of the federally threatened piping plover (Charadrius melodus), the federally threatened rufa red knot (Calidris canutus rufa) and the black-bellied plover (Pluvialis squatarola) and their prey in the context of stabilized and unstabilized shorelines in coastal Virginia. We evaluated 4 important foraging habitats (sandy ocean beach, peat banks, bay shoreline, and interior pools) on the 5 northernmost barrier islands in Virginia: partially stabilized Wallops Island and unstabilized Assawoman, Metompkin, Assateague Islands, and Gargatha Beach. In 2024 and 2025, we conducted monthly point count surveys from May 1st to August 31st for breeding season adult piping plovers and from April 15th to June 30th and August 1st to October 15th for migratory red knots, black-bellied plovers, and co-occurring shorebirds, and collected core samples to evaluate benthic prey availability. We also sampled terrestrial insects monthly from May to August due to their importance to piping plover foraging ecology during the breeding season. We implemented Bayesian mixed-effects modeling on a series of a priori candidate models to estimate the effect of distance to the nearest and each individual type of stabilization, along with important environmental and geomorphological covariates, on the presence of all three focal species and their benthic and terrestrial prey densities. We also compared our terrestrial prey densities and habitat distributions with historical data in Virginia and with 2 recent studies in New York and North Carolina to examine geographic or longitudinal differences in terrestrial prey availability for piping plovers in our study. We found that piping plovers were more likely to be present, and total benthic, worm and crustacean prey abundances were higher at greater distances from the stabilization actions during the piping plover breeding season at 4-5 years since the most recent activities. However, other factors such as habitat type and beach width were stronger predictors of piping plovers and their prey than distance to the nearest stabilization, with foraging piping plovers most likely to be present on peat bank habitats and on wider beaches. We found the opposite relationship with benthic insect larvae and terrestrial prey, which were more abundant closer to the stabilization actions during the piping plover breeding season; likely driven by their high abundances in the human-created borrow pool adjacent to the stabilization footprint. We found that red knot presence was more likely farther from the nearest stabilization features in all habitat types, but that stabilization was a weaker predictor of red knot presence than other ecological and geomorphic factors. Black-bellied plovers were strongly influenced by distance to stabilization, with contrasting relationships with stabilization depending on habitat type. Black-bellied plovers were more likely to be present farther from stabilized features when using peat, bay and pool habitats, all of which occurred outside of the footprint of stabilization. However, black-bellied plovers were more likely to be present closer to stabilization when foraging in sandy ocean intertidal habitat. Both red knots and black-bellied plovers were more likely to be present with higher abundances of other shorebirds and increased prey species richness, and black-bellied plovers were more present in areas where the shoreline change rate was negative (e.g., eroding or retreating). Our results showed that 4-5 years after shoreline stabilization actions on Wallops Island, benthic invertebrates in the sandy ocean intertidal zone had recovered to similar densities as the surrounding islands, but the highest densities occurred on peat banks outside stabilized areas. We suggest that future stabilization efforts both create high-quality foraging habitats, such as the borrow pool, within or adjacent to the stabilization footprint and conserve the potential for adjacent areas to respond to natural coastal processes that create bayside, pool, and peat habitats for foraging shorebirds. Future studies should deploy a before-after-control-impact design for at least 4 years pre- and post-action to better understand recovery times of avian and prey metrics, and to better tease out the impacts of individual stabilization actions.
- Of Hitlerization in Nazi Germany Between 1933-1938Elam, Kate (Virginia Tech, 2026-10-22)This thesis undertakes a theoretical analysis that applies the framework developed by Stephen Kotkin in his Stalinization as Civilization to the case study of Nazi Germany in the early 1930s. The process of social-engineering and indoctrination is analyzed and applied across two countries even if the strategies and tactics of such indoctrination differ to answer questions that have not been sufficiently explained about the rise of Adolph Hitler and his consolidation of power in the early years of the Nazi dictatorship. Like Kotkin, the thesis applies Foucault's concept of subjectivity to better understand power relations in Nazi Germany, specifically how an entire society of eager nationalists was created and largely underwritten by the public. This work fills a gap in the literature on the early years of the Nazi regime in that it attends in some detail to "the politics of everyday life," examining how individuals actively participated in shaping a new society relative to objectives that were only vaguely presented by the Nazi leadership. I use the Kotkin-Foucault concept of subterranean level of power to analyze the cultural dimensions of an emerging society in which individuals participated in morally heinous ethical and political acts through a set of predispositions to act in accordance with Nazi ideals of Aryanism. In the Soviet case, Kotkin argued that Stalinization created a state-led formation of societal followership through cultural, moral, and Marxist-Leninist values. I argue that Hitler was equally effective in creating a society of followership through a state-led social-engineering campaign called Hitlerization, in which individuals (starting with children) endeavored to "work towards the will of the Führer."
- Pseudoinverse Training for Neuromorphic Reservoir Computing Using Square-Root-Free Cholesky DecompositionLing, Jiaqiang (Virginia Tech, 2026-10-06)Adaptive reservoir readouts require a compact training engine as well as an efficient inference circuit. This thesis develops a fixed-point, square-root-free readout trainer for a windowed echo state network used in MIMO-OFDM symbol detection. The trainer streams regularized sufficient statistics into a banked, packed memory, factors the 48-by-48 normal matrix in place with a reciprocal-based LDL-transpose decomposition, and reuses the factors for all four real outputs. A word-length analysis of the datapath shows that the statistics multipliers can be narrowed from 48 by 48 bits to 14 by 14 and 14 by 17 bits without changing any result bit, reducing DSP usage from 162 to 36 slices on the Zynq ZCU104. For the factorization, a standard eigenvalue argument shows that ridge regularization bounds every reciprocal operand by the inverse of the regularization constant. This bound sets the required operand range and explains why a smaller 32-bit factor/solve option, which uses 26 DSP slices, passes tests with strong regularization but rejects every recorded packet it was tested on at the regularization used for detection. The selected design, implemented in the detection configuration, uses 36 DSP slices, 31,488 LUTs, and 27 block-RAM tiles, meets a fully constrained 50 MHz out-of-context timing requirement with an estimated 69 mW of dynamic power, and completes a 540-sample training batch in 147,146 cycles, or 2.943 ms. In RTL simulation of 19,230 recorded frames from 30 recordings in five scenarios, it trains every frame and reproduces all 3,692,160 reference weights exactly. With these weights, the ESN detector reaches a bit error rate of 4.05 percent over all frames, against 5.78 percent for LS/ZF and 4.90 percent for baseline MMSE detection, and is the better of the three in four of the five scenarios; pilot-tracked conventional detectors remain more accurate. A triplet-STDP circuit study is also summarized, and the adaptation of the trainer to GlobalFoundries 40LP is identified as ongoing work.
- Experimental Investigation of Two-Bubble Cavitation Collapse Near a Rigid Wall: Effects of Stand-Off and Inter-Bubble DistanceUpadhyay, Kushal Jitendrabhai (Virginia Tech, 2026-10-16)Cavitation bubble collapse near solid boundaries can generate intense pressure waves and liquid microjets that contribute to material erosion and structural damage. This study experimentally investigates the interaction and collapse of two laser-induced cavitation bubbles arranged coaxially near a rigid wall, focusing on the effects of bubble-to-wall stand-off distance and inter-bubble separation. Two bubbles were generated in deionized water using independently controlled pulsed Nd:YAG lasers. High-speed shadowgraphy and Schlieren imaging were used to characterize bubble dynamics and pressure-wave propagation, while a flush-mounted pressure sensor measured the transient wall pressure. The stand-off distance was varied from γ = 0.52 to 1.97, and the inter-bubble distance was varied between β = 5.84, 4.92, and 4.03 mm. The results show that stand-off distance strongly affects collapse morphology and wall pressure. At small stand-off distances, both bubbles undergo toroidal collapse, while at larger distances the near-wall bubble transitions to tip collapse. The maximum first collapse wall pressure of approximately 34.8 MPa was measured at γ = 0.52 and β = 5.84 mm. The farther bubble generally collapses first, generating a pressure disturbance that modifies the collapse of the near-wall bubble. Reducing the inter-bubble distance does not monotonically increase wall pressure; instead, the pressure response depends on bubble interaction, collapse timing, deformation, and energy redistribution. These findings demonstrate the importance of bubble-bubble and bubble-wall interactions in determining cavitation-induced wall loading.
- Using Nacre-Inspired Architecture to Improve the Mechanical Performance of Binder Jetted PartsEspinosa Borja, Juan Fernando (Virginia Tech, 2026-10-15)Residual porosity remains a barrier for industrial adoption of additively manufactured metallic parts fabricated through Binder Jetting (BJT), due to its detriment in the mechanical performance. However, local control of the pore structure poses an opportunity to turn an apparent process defect into a design feature for creating architected composites. Nacre is a natural composite that achieves high toughness by directing crack propagation through a softer, more compliant phase, effectively preventing failure in the brittle ceramic material. This research investigates whether a nacre-inspired architecture can be introduced into BJT metal composites to improve toughness relative to conventional, homogeneous parts. Two strategies within a BJT and infiltration workflow were used to control porosity at different length scales: selective binder deposition governs mesoscale porosity by dictating where powder is bound or left loose during printing, while pre-sintering provides control of the microscale porosity by controlling neck formation and pore evolution within each region. This work leverages both controls together to architect distinct infiltration pathways within a single part, producing a Nacre-inspired brick-and-mortar pattern of bronze-lean "Brick" (unbound) and bronze-rich "Mortar" (bound) regions in SS316L and bronze composites. The progression of densification of bound and unbound SS316L powder throughout the sintering process was found to differ, with unbound powder showing an advanced densification because of its greater powder packing density. A pre-sintering temperature of 1250℃ produced the greatest contrast in pore structure while providing a viable, complete infiltration path. Nacre composites were fabricated at this temperature, achieving up to a 9% bronze-content difference between the designed Brick and Mortar sections. Conventionally printed and infiltrated composites outperformed the Nacre composites, with Modulus of Toughness recording values of 0.024 J/mm3 and 0.016 J/mm3, respectively, with stiffness showing no distinction. These results are attributed to large fabrication defects and high residual porosity in the final parts (6.6 – 8.5%), rather than to the tailored architecture itself. This thesis demonstrates a novel method for producing complex architected composites with spatial composition contrast in a BJT and infiltration process, by combining selective binder deposition and a pre-sintering step. Additionally, it develops an image-analysis method applicable to other porosity-based patterns for quantifying such contrast from optical microscopy.
- Predicting Heat Accumulation and Layer Geometry Variation in Continuous DED-Arc of Thin-Walled Aluminum StructuresRosenhein, Drew David (Virginia Tech, 2026-10-15)Compared to other metal additive manufacturing (AM) processes, wire-arc directed energy deposition-arc (DED-Arc) offers high deposition rates. However, its throughput is limited by heat accumulation within the part during the build; continuous printing without balanced heat dissipation can lead to part slumping and distortion. Existing thermal mitigation approaches, including interlayer dwell time, active cooling, and in-process parameter control, impose costs in productivity, system complexity, and/or limited applicability to large and complex parts. Continuous deposition without arc interruption eliminates dwell time and arc start and stop defects but produces a severe heat accumulation condition. Under fixed process parameters, the thermal balance determining this accumulation is strongly influenced by the part's geometry. In a closed contour geometry, the layer contour length of the profile determines the toolpath length, setting the time available for heat to dissipate between depositions of successive layers, while also determining wall surface area. Longer layer contours have been observed to reduce heat accumulation in previous research, but the relationship has not been directly investigated as a design variable in DED-Arc, nor has it been established whether interlayer temperature reaches a plateau at sufficient contour length and/or part height. The aim of this thesis is to establish a design-stage workflow that relates continuous build geometry to its in-process thermal condition and deposited wall geometry, thus allowing layer-wise prediction of part geometry before fabrication. To close this research gap, three thin-walled ER5556 aluminum cylinders of 50, 100, and 200 mm diameter were deposited continuously under identical process parameters and instrumented with embedded thermocouples along the build height. Contour length variation produced a 212.1 °C spread in interlayer temperature across builds and dictated whether steady-state interlayer temperatures could be reached. Calibrated thermal simulation indicated that the interlayer temperature of 200mm diameter cylinder reached a plateau during processing and declined as convective loss exceeded layer-wise heat input. However, the 50mm diameter contour part's interlayer temperature climbed to 401 °C by layer 50 with no sign of stabilizing. This thermal variation increased effective wall width by 99.4% along the part height, whereas porosity and microhardness remained consistent for this alloy. The calibrated simulation was applied to a test geometry and, combined with the measured temperature-geometry correlations, predicted effective wall width and layer height to 3.28% and 2.26% mean absolute percentage error, respectively, across 10 layers sampled from a 50-layer build. The study was also extended to oscillating weave deposition for producing thicker wall widths. Interlayer temperatures in the weave builds were up to 60.4 °C lower than single-track at matched contour length, despite depositing 2.8 times the path length per layer. This longer layer time increased heat dissipation, more than offsetting the higher heat input.
- Through The Keyhole: A Nature RetreatPerry, Mason Thomas (Virginia Tech, 2026-10-15)Looking Through the Keyhole: A Nature Retreat explores how architecture can encourage meaningful social connection while providing opportunities for privacy, reflection, and reconnection with nature. The project responds to the growing separation between people, community, and nature by proposing a retreat organized around spatial transitions rather than isolated buildings. The design consists of three primary architectural moments: the threshold gate, firepit, and elevated cabins. The threshold transitions from everyday life to the retreat, the firepit creates a communal center, and the elevated cabins offer increasingly private spaces. A recurring keyhole motif symbolizes transitions between public and private, exterior and interior, and movement and reflection. These elements create a progression from crossing, gathering, to ascending, demonstrating how spatial sequence, material expression, and nature connection shape social interaction, solitude, and restoration.
- Communication Strategies for Cognitive Radar NetworksEsser, Joseph Max (Virginia Tech, 2026-10-14)Distributed radar networks provide the ability to monitor large and dynamic scenes by uti- lizing multiple spatially separated sensors. However, the effectiveness of such networks are constrained by limited communication resources if a central fusion center needs to collect raw data from all the radar nodes. This thesis investigates adaptive radar resource manage- ment for multi-radar multi-target tracking through the application of multi-armed bandit based scheduling methods. A simulated radar network is developed where multiple radars observe maneuvering targets while providing track updates to a centralized fusion center. Several radar selection strategies based on ϵ-Greedy and Upper Confidence Bound (UCB) multi-armed bandits are investigated and compared against a baseline communication strat- egy. Radar selection is performed by utilizing contextual information at the fusion center derived from the current tracking state and spatial distribution of targets, which allows communication resources to be dedicated to radars of interest. The performance of the re- sulting developed tracking system is evaluated using the Generalized Optimal Sub-Pattern Assignment (GOSPA) metric. Simulation results demonstrate that the adaptive scheduling methods improve tracking performance by concentrating resources on regions of high target density, while the relative results between communication methods depends highly on target conditions and communication constraints.
- Influence-Guided Agent Adaptation for Dynamic Human–AI CollaborationDubey, Sara (Virginia Tech, 2026-10-12)To facilitate effective human–agent collaboration, agents must not only understand the intent of their human partners and adapt to it reactively, but also account for how their own behavior influences their partners. Rather than treating this influence as a passive consequence of interaction, we propose actively predicting and optimizing it as part of the agent's decision-making process. In doing so, the agent explicitly models how its actions are likely to shape the subsequent behavior of its human partner. To this end, we introduce i-STAR, a novel framework that not only optimizes an agent's own actions but also reasons over and jointly optimizes the anticipated response of the human partner to those actions. Our framework employs a multi-layer architecture comprising three core components: (1) a team response model, which predicts how the human partner will respond to candidate agent actions; (2) a joint utility model, which estimates the overall utility of the agent's selected action together with the anticipated partner response; and (3) an action policy whose advantage function is shaped by the maximum-utility joint action. We find that, using this framework, our agent outperforms current state-of-the-art agents across all layouts of the challenging Overcooked benchmark. Furthermore, in a human-subject study, i-STAR is perceived as significantly more fluent than the strongest agent–agent baseline.
- Failure Analysis and Re-Design of an Electromagnetic Energy Harvesting Railroad TieJadhav, Advait Vishal (Virginia Tech, 2026-10-08)Advanced Rail Technologies call for intelligent and autonomous trackside monitoring, signalling and communication equipment. This equipment must be powered where it is installed, and across much of the rail network no power supply exists. The Energy Harvesting Tie developed at the Center for Vehicle Systems and Safety at Virginia Tech captures the vertical deflection of the rail under passing wheels within the enclosure of a standard composite crosstie, converting it through a ball screw and bevel gear stage into unidirectional generator rotation by half-wave mechanical motion rectification. Following instrumented laboratory and revenue-service campaigns, the first-generation prototype was left in track to expose degradation mechanisms inaccessible to short-term testing. At nine months of continuous service, it ceased to produce voltage on any generator phase, abruptly and with no preceding decline. This thesis establishes the cause and develops a second-generation design. Teardown of the recovered unit and a failure modes and effects analysis identify degradation of the one-way clutch cage and sprags as the dominant risk driver, and reconstruct a causal chain of lost sprag positioning, lubrication loss and internally generated debris, axial migration against a compliant polymer locating member, and diversion of transmitted torque into a parasitic frictional path at the gearhead adapter joint. Eight design requirements follow, each traceable to a specific teardown observation. Design loads are derived from the revenue-service displacement record at the electrical operating point imposing the greatest mechanical duty. A nodal torque-propagation model retaining every rotating inertia gives a peak clutch torque of 23.02 N·m at zero stress ratio, an indexing rate of 0.7-1.5 Hz, and a ball-screw axial reaction reaching 11.58 kN and shows that the conventional reduction of freewheeling inertance to the generator rotor and gearhead alone understates it significantly. The redesign mitigates issues concerning race concentricity, axial location and gearhead torque reaction by consolidating the components, using a complete sealed clutch unit located by co-rotating metallic shoulders and a two-point supported eight-fastener generator mount in place of a cantilevered four-fastener arrangement. Every load-bearing member is verified analytically. Testing of the assembled unit over 131 compression events confirms half-wave rectification on consecutive events, validates the drivetrain kinematic chain, and recovers the machine constant in situ, with performance holding up to expectation. Instrumented revenue-service redeployment remains the necessary next step.
- Acoustic Sensing and Asymmetry: An Improved Model for Hearing Mechanics in Ormia ochracea Or, When 'Hearing Crickets' Is a Good ThingMikel-Stites, Maxwell (Virginia Tech, 2026-10-08)Ormia ochracea is a parasitoid fly known for its ability to precisely locate the origin of sounds despite its small size. It achieves this via a hearing organ composed of a pair of external tympanal membranes joined by a biomechanical bridge. The fly has inspired research projects ranging from novel micro-scale acoustic sensors to the biomechanics of hearing. However, there has been relatively little work re-examining the physical measurements of O. ochracea's sensory organs, nor the original mathematical model (a system of coupled ODEs) for ormiine hearing (first established in the mid 1990s). In the case of the former, further study provides a chance to verify the physical measurements in the literature and explore the possibility of physical asymmetry in ormiine hearing. For the latter, further work would address significant deviations between the measured and predicted sensory values relevant to ormiine hearing. To address these gaps, in our first study, we made morphometric measurements of 21 Ormia specimens and quantified the degree of tympanal asymmetry in order to determine whether the system may leverage small biologically deliberate asymmetry for a sensory advantage. In our second, we modified the prior model by adding a response to the angle of incoming sound, improving its accuracy and overall fidelity to physiological measurements when compared to the unaltered model. Together, these studies reveal the prevalence and impact of tympanal asymmetry in O. ochracea and suggest avenues of improvement for both further mathematical modeling and experimental micro-scale physical experimentation.
- Exploring the Impact of Institutional Measures on Household Disaster Preparedness through a Recreancy Lens: A Case Study of Hurricane Beryl in TexasKona, Afsana (Virginia Tech, 2026-10-08)This study uses recreancy as a theoretical and analytical hook to explore how institutional measures affect household disaster preparedness. As the concept of recreancy is broadly linked to institutional failure during a crisis, this study focuses on Hurricane Beryl (2024) in Texas as a case study. The study examines how risk perceptions, institutional trust, disaster experience, and information reliance shape household-level preparedness behaviors. The study deploys a survey research design to collect data from 540 respondents across 23 counties affected by the deadly Hurricane. For quantitative analysis, the study used Stata 18.5 version of the software and conducted descriptive analysis as well as ran hierarchical ordinary least squares regression models. Responses to open-ended questions were analyzed thematically to contextualize and interpret participants' insights. Quantitative results suggest that higher risk perception and prior disaster experience are positively associated with preparedness behaviors. However, the relationship between institutional trust and perceived institutional failure is complex and does not follow a linear pattern. While trust can facilitate preparedness, it may also reduce perceived urgency to act. In contrast, perceptions of institutional failure are associated with reduced reliance on formal information sources and, in some cases, lower preparedness. Analysis of open-ended responses further highlights challenges from inconsistent communication and dissatisfaction with prior institutional responses that adversely affect the community's preparedness actions. This study contributes to the ongoing discourse on disaster and society by emphasizing that preparedness is not solely an individual behavior but is deeply influenced by institutional credibility and performance.
- Formulation and Microstructural Evaluation of Multi-Component Rare Earth Oxide Thermal Barrier Coatings via Slurry DepositionMassignan, Connor James (Virginia Tech, 2026-10-08)Thermal/Environmental Barrier Coatings (T/EBCs), composed of multi-component rare earth oxides (REOs), offer tunable properties for high-temperature applications due to their low thermal conductivity, high melting temperature, and low coefficient of thermal expansion. Slurry deposition of coatings is a cost-effective alternative to state-of-the-art air plasma spray methods. This study investigates a ternary REO system comprising yttria (Y2O3), erbia (Er2O3), and ytterbia (Yb2O3), to assess its suitability as a single-layer TBC material. Slip-cast samples were fabricated at varying volume percentages and sintered under controlled conditions to explore the influence of particle packing and sintering parameters on densification behavior. Then, the identified conditions were used to slurry spray a niobium-based alloy. Emphasis was placed on optimizing the slurry formulation to systematically investigate varying stabilization mechanisms to develop the suspension, influencing the rheological behavior and particle packing of components consisting of slip-cast and slurry-spray coatings. Preliminary results suggest that sintering environment and temperature significantly affect final density and coating integrity, as well as the percentage of yttria, warranting further investigation into the thermophysical interactions among REO constituents during thermal treatment.
- Studies on Particle-Wall Interactions in Particle-based High Temperature Concentrating Solar Thermal SystemsOwens, Conner David (Virginia Tech, 2026-10-08)Next-generation (Gen3) concentrating solar power (CSP) systems use solid particles as the heat capture, transfer, and storage medium to reach operating temperatures above 700 °C, but the continuous contact between flowing particles and component surfaces drives severe material degradation through the coupled action of high-temperature oxidation and abrasion. This thesis examines particle–wall interactions in particle-based high-temperature CSP through two complementary studies. First, the combined oxidation–abrasion performance of four candidate alloys, stainless steel 316 (SS316), Incoloy 800H, Haynes 230, and Inconel 718 (In718), is evaluated in a moving bed of dense sintered bauxite particles (CARBO HSP 40/70) at room temperature, 700 °C, and 800 °C. The iron-based alloys exhibit relatively stable abrasion behavior across the temperature range, whereas the Ni-based alloys undergo intermittent severe degradation at 700 °C but improved long-term wear resistance at 800 °C, where enhanced oxide regeneration and mechanically mixed layer development provide increased surface protection. Second, a nanoscale multilayered Cr/Cr2N coating deposited on In718 by reactive unbalanced DC magnetron sputtering is tested against uncoated In718 for 324 h at 700 °C and 800 °C under combined oxidation and abrasion. The coated specimens show no measurable substrate mass loss at either temperature, while uncoated In718 at 700 °C spalls at an average rate of 4.47 mg/h, identifying nanoscale Cr/Cr2N multilayers as a viable route to extending component service life. Collectively, these studies indicate that a self-replenishing mechanically mixed layer forms naturally under sustained particle contact and provides an additional protective barrier against abrasion, provided the underlying substrate or coating is capable of maintaining oxide-scale adhesion and mechanical stability under dynamic particle contact. Because the alloy is then not directly exposed to the abrasive medium, this behavior theoretically allows lower-cost materials to be used in particle-handling components, and provides a basis for the design of longer-lived, more economical receivers for Gen3 CSP.
- A meta-analysis on the relationships of agency and communion with job strainKalmar, Sam L. (Virginia Tech, 2026-08-31)Strain is a common phenomenon experienced in the workplace that can induce serious negative consequences for those who let it go unchecked. Individual differences are known to play distinct roles in the severity of these outcomes, but the limited exploration of personality traits has yielded inconsistent findings. The current study helps close this knowledge gap by examining the relationship that the personality traits of agency and communion exhibit with job strain outcomes. Approximately 11,496 citations were screened, resulting in 290 unique independent samples, 563 unique effect sizes, and 67,564 participants. A random-effects, individual correction meta-analysis was conducted for both the agency-strain and communion-strain relationships. Continuous moderators were analyzed with mixed-effect meta-regressions and categorial moderators were analyzed with modified asymptotic confidence intervals. Both agency and communion had negative relationships with job strain outcomes, although agency’s relationship was stronger than communion’s. Occupational orientation and sample age did not appear to moderate these relationships. In sum, both agency and communion appear to mitigate job strain, yet agency seems to be more effective overall. Further research is needed to determine if methodological moderators such as strain and personality measure types influence these relationships.