Research · The program

Models that read rooms.

Research at Spatial Web is applied: it exists because the corpus exists. Five operating companies produce photoreal, measured, labeled interiors every day — the program studies what machines can learn from them. Papers publish here first; drafts circulate earlier by request.

Papers
SW-26.01
2026

The Residential Corpus

Documentation paper

What listing capture actually contains: photoreal imagery, lidar-derived geometry, listing-derived semantics, and verified market records — assembled as a by-product of operating real estate companies. Covers scale, formats, provenance, and rights, and argues that lived-in homes are the missing distribution in physical-AI training data.

In preparationRequest draft
SW-26.02
2026

Spatial LLMs: Models That Read Rooms

Position paper

What changes when language models are grounded in the built environment: tokenizing space, aligning geometry with description, and evaluation beyond captioning. Sets out the architecture and data requirements — and the case that market-grade metadata outperforms crowd labels.

Forthcoming
SW-26.03
2026

From Listing to Label

Methods paper

Market records as free ground truth: using ownership, valuation, and listing metadata to supervise indoor scene understanding without an annotation pass. Every home arrives pre-labeled by the transaction that sold it.

Forthcoming
The corpus, specified

One capture, four layers. Each environment in the corpus carries all four — produced by commerce, joined at the address.

L1Imagery
Photoreal capture of lived-in interiors — true lighting, real furnishings, real materials in place.
L2Geometry
Lidar-derived structure: meshes, floor plans, dimensions, walkable space.
L3Semantics
Listing-derived labels: rooms, beds, baths, square footage, finishes — ground truth without an annotation pass.
L4Market record
Verified per-address metadata: ownership, valuation, transaction history.
Scale
~50,000 residential environments, growing daily.
Delivery
USD / glTF — Isaac Sim and Omniverse-ready.
Refresh
Continuous — every new listing extends the corpus.
Rights & provenance
Documented in SW-26.01.

Fig. 1 — spec sheet, abridged. Full specification ships with SW-26.01.

Access