Skip to content
DocsHow indoor navigation works

DocsGetting started

How indoor navigation works

From the organization’s map to a visitor’s directions: locations, connections, reference images, camera localization and the route.

On this page

Veeee has no beacons, no floor-plan tracing and no GPS. Three things make navigation possible: a graph of locations and connections, reference images for each location, and a camera check that compares what a visitor sees with those images.

1. The building is a graph

Each indoor map covers one building. Inside it, every place a visitor can be is a location, and every walkable link between two places is a connection with a distance in metres. Floors are a number on each location, so a connection that joins Floor 0 to Floor 1 is a stair or an elevator. Routes are calculated over this graph: the route planner finds the shortest path by distance, weighted by each connection’s traversal cost, and skips connections that are temporarily blocked.

Mapped turns can also be recorded as hidden navigation geometry: points that shape the drawn route and the turn instructions but are never offered to visitors as destinations.

2. Reference images teach Veeee each location

For every location, the organization adds photos or a short video. Veeee processes them into a visual index. A location is ready for camera recognition once it has enough usable images; the workspace shows the minimum (three by default) and which locations still need more. Permanent features such as doors, signs, counters and corridor intersections work best. See Reference images.

3. The camera finds the visitor

When a visitor taps Start navigation, the browser opens the rear camera. Two small frames are captured a fraction of a second apart and sent together to Veeee, which compares them with the map’s reference images. The location is accepted when both frames agree or one frame is a strong match; if the view is ambiguous the visitor is asked to show a door, a sign or a wider angle, and the check repeats automatically. The frames are downscaled before sending and are used only for this comparison.

A location QR code can replace this first check: a link with a start point tells Veeee where the visitor is, and the camera is used from then on to keep track. See Current-location recognition.

4. The route and the instructions

With a start and a destination, Veeee plans the route and starts a navigation session. While the visitor walks, the camera keeps checking their position against the map, and the phone’s motion sensors estimate footsteps and heading on the device. Instructions are written as a person would say them: “Turn left in about 8 m”, “Take the elevator to Floor 2”, “Room 210 is on the left”. If the visitor leaves the route, Veeee recalculates it from the confirmed position. See Route instructions.