Lumina

Home energy assessments are bottlenecked by the survey step: a trained assessor visits each home, measures rooms by hand, photographs construction details and manually enters everything into a simulation tool. It's slow, expensive and introduces transcription errors at every step. Lumina's premise is that a modern iPhone has everything needed to automate the physical measurement part of that workflow.

The app uses ARKit and the iPhone's LiDAR scanner to build a point cloud of each room in real time. A geometry extraction pass identifies walls, floors, ceilings, windows and doors from the point cloud — without requiring the assessor to label anything. Room dimensions, ceiling heights and opening sizes are computed automatically and immediately available to the model back-end.

Feature detection

LiDAR gives you geometry; it doesn't give you construction type. Lumina combines the geometry with a classification step that uses the depth map and RGB frames together: window glazing type, wall finish and (with some probability) insulation presence can all be inferred from surface texture and thermal gradient signatures visible on the point cloud. Where the classifier is uncertain, the app prompts the assessor for a manual tag — so the automation handles the easy cases and the human handles the hard ones.

Live sync

As rooms are scanned, measurements stream over a WebSocket connection to the thermal model back-end (built on the Enthalpy engine). The assessor sees the model's predicted heat loss and energy demand update in real time as they move through the house — turning the survey from a data-collection exercise into an interactive energy audit.

Status

Lumina has a working scan-and-extract pipeline and the WebSocket sync layer. The classifier and the live model integration are under active development. It's closely coupled to the Enthalpy project — the two are designed as a system.