Dec 2025 // Present

Developing Synchronized Georeferencing Methodology

Digital twins are interactive 3D models of built environments that support facility planning, wayfinding, virtual tours, and long-term spatial documentation. Yet, most workflows depend on costly LiDAR scanners, paywalled software, and specialized processing that put them out of reach for small institutions and commercial operators. Structure-from-Motion (SfM) photogrammetry paired with 3D Gaussian Splatting (3DGS) offers a low-cost alternative capable of photorealistic, real-time rendering in a web browser. However, interiors lack GNSS coverage, so indoor and outdoor spaces are typically modeled separately in unrelated coordinate frames. Introducing Inside Out, a project with two linked objectives. The primary objective is an accessible, repeatable methodology for modeling building exteriors and interiors. Exterior imagery is collected with a consumer drone (DJI Mini 4 Pro) flying standardized nadir and oblique plans over GNSS-surveyed ground control points. Interior spaces are captured with 360° cameras on a multi-height pole rig, and the equirectangular video is reprojected into SfM-ready rectilinear views using FFmpeg. All processing software is free, and total hardware costs roughly $4,000. The secondary objective builds on the first: tethering interior and exterior models into a single georeferenced digital twin. AprilTag fiducials placed at doorways serve as shared control, carrying exterior scale and position into GNSS-denied interiors so every room shares one coordinate frame. Applied to a rural church and to Henson Science Hall at Salisbury University, the workflow produces visually coherent splats with legible façades, rooflines, and interior detail. Inside Out lowers the barriers to producing photorealistic, connected digital twins.

Oct 2025 // Present

Paused Until Inside Out's Completion

Traditional drone photogrammetry can produce noticeably different results under different lighting and shadow conditions, even when flight plans and camera settings stay the same. As a research subproject within the broader Salisbury University Digital Twin effort, Chasing Shadows tests whether “illumination diversity” can improve small-UAS Structure-from-Motion (SfM) reconstruction quality without adding expensive sensors or complex field workflows. I fly the same site three times per day (morning, solar noon, and late afternoon) (all flights stay in civil daylight) while holding altitude, overlap, and camera settings constant so lighting becomes the primary variable. I process each dataset independently in RealityScan and compare outputs against control points to quantify RMSE, bias, and dispersion, then evaluate whether selective use or fusion of datasets improves accuracy and stability. This work feeds directly into the larger campus modeling pipeline by informing when and how we capture exterior imagery, helping the digital twin team standardize missions that produce more repeatable, higher-quality reconstructions at the lowest practical cost and complexity.

May 2026 // June 2026

Most of my fieldwork communication depends on cellular coverage or a single point-to-point radio link. This project tests an alternative: a decentralized LoRa mesh network that relays messages and location data between nodes with no cell service or internet connection required. The node runs Meshtastic, open-source mesh firmware, on an off-the-shelf ESP32 + LoRa development board, paired with a custom enclosure I designed in Onshape and FDM printed for portability. Setup includes flashing firmware, configuring the device for the US 915 MHz ISM band, and testing range and multi-hop relay performance against a single point-to-point link. The goal is twofold: build hands-on fluency in embedded systems and RF fundamentals that commercial mesh hardware abstracts away, and evaluate whether a low-cost mesh node could support communication during remote capture work in areas where cell coverage is unreliable or absent.

Nov 2025 // Dec 2025

Motor vehicle crashes remain a major public health concern in the United States; recent estimates indicate more than 42,000 fatalities in 2022 and a national fatality rate around 1.3 deaths per 100 million vehicle miles traveled. This project uses a geographic information system (GIS) to identify high-risk road segments in Wicomico County, Maryland, and to quantify the exposure of nearby residents, public K12 students, and Shore Transit bus routes. Average annual daily traffic (AADT) data for 2010 to 2018 from the Maryland Department of Transportation State Highway Administration (MDOT SHA) were integrated with 2019 to 2023 crash locations from the Maryland State Police crash data download portal. Additional layers include 2020 Census blocks, 2024 to 2025 public school locations with 2023 to 2024 enrollment, and May 2025 Shore Transit routes.

A "crash rate" was computed for each road segment as crashes divided by the product of AADT and segment length in kilometers. Segments with crash rates greater than 0.05 were defined as significant risk segments, and a 1 km buffer around these segments delineated “crash risk zones”. These zones were intersected with Census blocks (using area-weighted interpolation), school locations and enrollment, and Shore Transit routes. Results show that high-risk segments are concentrated in and around Salisbury and along US-50 and US-13. Approximately 27,072 residents, about 26% of the county’s 2020 population, live within 1 km of a high-risk segment. 5,175 public K12 students (33% of county enrollment) attend schools inside the crash risk zones, and 45.5% of Shore Transit route mileage within Wicomico County lies in these zones. The study illustrates how relatively simple GIS operations can reveal how roadway risk intersects with people, schools, and transit, while also highlighting limitations related to data vintages, areal interpolation, and the modifiable areal unit problem.

Sep 2025 // Nov 2025 (Maintenance Stopped Spring 2026)

This project focuses on transforming Salisbury University’s main Commons dining hall and its satellite/pop-up dining locations into an interactive digital experience. Using ArcGIS Pro, Field Maps, and ArcGIS Online, I am mapping and digitizing all dining service locations, creating a web-based platform for students, faculty, and visitors to easily explore, locate, and interact with each site.

Though carried out independently, the project has been formally approved by the Department of Geography & Geosciences w/faculty offering support as needed. The end product will provide an accessible GIS-powered tool that blends cartography with web interactivity, enhancing campus engagement and decision-making for both the university and its community.

While the project was completed in November 2025, ongoing updates being done by me.

Jun 2025 // Aug 2025

Project explored how DEM and LiDAR datasets can be transformed into physical 3D-printed models for landform and watershed analysis. Using CAD workflows and slicing software, I processed raw spatial data into printer-ready files that accurately represent topography and hydrological features.

Dec 2024 // Aug 2025 (Ongoing Maintenance)

The Norrisville United Methodist Church graveyard holds a rich history and serves as the final resting place for numerous individuals, including veterans who served our country. This project seeks to preserve and make this history accessible through the application of cutting-edge technology. Using drone-based photogrammetry, I will create a detailed 2D & 3D digital models of the entire graveyard, capturing the location and condition of each headstone. This model will then be integrated with GPS data to accurately pinpoint the location of each grave. The final deliverable will be an interactive platform allowing users to virtually explore the cemetery, locate specific graves, and access information about the individuals buried there, with a particular focus on providing details about the veterans interred. This project aims to enhance community engagement with local history, facilitate genealogical research, and provide a lasting digital archive of this important site.

Sep 2023 // May 2024

Between June 23rd, 1851, and December 31st, 2022, 606 recorded hurricane landfalls have occurred on United States’ soil, spanning from the Gulf of Mexico to the Northern Atlantic Basin. Utilizing Geographic Information Science (GIS), multiple layouts were created via a series of analyses regarding elevation, population, and ocean water temperature data. It was predicted that the frequency of hurricanes, average sea temperatures, and the magnitude of hurricanes including their respective impact on populated areas, would increase over time. The study showed that the number, frequency, and severity of hurricane landfalls increased from June 23rd, 1851, to December 31st, 2022. In addition, the most likely states to encounter a hurricane landfall were Florida, Louisiana, Texas, South Carolina, and North Carolina. It should be noted that the 2010 to 2019 decade had greater temperature anomalies compared to the 2000 to 2009 decade. Areas that were most likely to experience landfall tended to have the lowest elevations, increasing the risk of catastrophic flood-related damages. Critical coastal population centers were resilient for one to three meters of inland storm surge. However, once inland storm surge peaked at five to nine meters (worst case scenario), these population centers were very susceptible and there were cases where the water was trapped inside of levee systems, increasing impact. Factors that may explain these results include the influences that changing ocean and wind currents have on the environment, the curvature of the U.S. coastline, changing sea temperature in areas relative to each landfall/ocean section, and an increase in population over time.