For the complete documentation index, see llms.txt. This page is also available as Markdown.

App Localization

To try out localization in the app, click on the Map name card after the map status is active

Choosing a Localization Type

Selecting a Map or MapSet opens the Start Localization dialog, which offers two ways to localize. Both return the same 6-DoF pose, they differ in how many camera frames are sent and how long the request takes.

Start Localization dialog offering Robust Localization using multiple frames and Fast Localization using a single frame
The Start Localization dialog, shown after you select a Map or MapSet.

Robust Localization (multiple frames)

Captures 4 to 6 frames along with the device's ARKit or ARCore tracking pose and sends them together. Combining several viewpoints makes the result more reliable in difficult conditions such as repetitive textures, partial occlusion, or minor changes to the space since it was scanned. Expect a pose in up to around 5 seconds.

Fast Localization (single frame)

Sends a single camera frame. This is the quicker option, returning a pose in around 2 seconds, and works well in well-textured, stable spaces where you want a result immediately.

Robust Localization
Fast Localization

Frames per request

4 to 6

1

Typical response time

up to ~5 seconds

~2 seconds

Best for

Challenging or visually repetitive scenes

Quick results in stable, feature-rich spaces

These correspond to the multi frame and single frame query types in the VPS API. See Localization for the underlying endpoints and parameters.

Running the Localization

Once you pick a localization type, the camera will open to initiate the localization process. Move your phone left and right within the mapped area to allow the camera to perform accurate localization.

After successful localization, you will see your Map mesh aligned accurately with the real world, as shown below.

Last updated

Was this helpful?