Agent-Based Digital Twin Simulation
May 2026 - Present
Throughout the past couple of months, I've been working with the GeoHDR research team at UNC Charlotte to create an ABM (Agent Based Modeling) framework in the Unity Engine. This framework is capable of simulating 800+ student agents in a 3D replica of a real school building. The project's goal is to combine the ABM framework with a deep learning airflow surrogate, allowing us to simulate disease transmission in indoor multi-floor scenarios, without the need of actual human test subjects.
Digital replica of the McEniry building at UNC Charlotte
Real picture of the McEniry building
Agents are able to navigate all four floors of the building using stairs or elevators. They also generate their own class schedules using real course data. Using the ABM framework, Agents can achieve realistic behaviors such as going to the bathroom, studying, heading to office hours and more.
Agents navigating the hallways of McEniry
Every agent has their own schedule (top right). It also displays their activity status.
