UC Berkeley Campus Digital Twin

The campus environment is created from drone imagery, LiDAR scans, photogrammetry and Gaussian Splatting models, GIS building data, BIM models, and manually prepared assets. Each model is placed within its correct campus location so buildings, roads, terrain, and infrastructure remain spatially connected.

ArcGIS Pro prepares and aligns geospatial information. Blender supports cleanup, editing, and visual refinement. Unreal Engine brings the final assets together as an interactive environment. Different levels of detail keep important buildings visually rich while allowing the full campus scene to remain responsive.

The intelligent copilot helps users explore campus information without needing to know where every report or dataset is stored. Users can ask questions about energy, water use, building information, or other available documents directly inside the digital twin.

The goal is a more accessible interface between people and technical information. Students, visitors, engineers, and decision makers can interact with the same campus data at the level of detail they need.

When a user selects a building, the platform can display its construction year, floor area, primary use, history, and heating or cooling systems.

This connects the physical 3D model with information that would normally be stored in separate reports, websites, or facility databases. Users can explore the campus and access information directly from the location it describes.

The feature can support visitor orientation, engineering education, facility review, and future maintenance planning.

Ground movement can be difficult to understand when it is shown only as a flat map, color scale, or table of values. This module brings InSAR displacement information and subsurface models into the 3D campus environment.

Users can explore how movement changes over time and see where displacement occurs in relation to buildings, roads, utilities, slopes, and other campus assets. Subsurface information can also be represented as 3D blocks showing different geological or material conditions below the ground.

The module creates a foundation for future infrastructure monitoring, ground risk assessment, and maintenance planning.

The fire response module connects a possible building incident to the campus water network. It helps users identify nearby hydrants and understand how each hydrant relates to surrounding pipes.

The recommendation process can consider distance, water pressure, flow capacity, network connectivity, valve status, and hydrant availability. This provides more useful information than simply selecting the closest hydrant on a map.

The selected hydrant and related pipe network can be viewed in both the 3D environment and campus map, giving emergency teams clearer information about location, access, and network conditions.

Digital twins can support remote inspection of tunnels, underground systems, and confined spaces that may be difficult or unsafe to enter. The tunnel shown below is one example of how a real environment can be documented and reviewed from another location.

I created its 3D model from reality capture data, brought it into Unreal Engine, and developed it as a VR experience so the remote review feels more immersive and spatially intuitive.

Engineers and students can explore visible conditions, discuss areas of concern, and prepare inspection plans before entering the site. This approach can reduce unnecessary exposure while making technical review and training more accessible.