Simultaneous localization and mapping, or SLAM, estimates a device's position while building or updating a map of its surroundings.
Two questions answered together
The system needs to know where it is to map what it sees, but it also needs a map to work out where it is. SLAM addresses these linked problems by combining observations over time. Visual-inertial systems can combine camera information with motion sensors.
The map may not look like a room model
A tracking map may contain features useful for localisation without being a detailed, textured mesh that an artist can export. Tracking a space and producing a reusable 3D asset are different goals.
An exhibition example
A system may estimate how a visitor's device moves around a virtual object placed in a room. That supports spatial consistency while the experience is running.
Practical limits
Low-detail surfaces, poor light and changing surroundings can make localisation harder. Drift and tracking loss are possible. SLAM alone does not promise shared experiences, exact measurements or placement that survives closing and reopening the browser.