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

Frontier Exploration

Autonomously explore and map an unknown space during 2D SLAM.

Mapping a space by hand — teleoperating the robot into every corner — is tedious. Frontier exploration does it for you: the robot heads for the edges between mapped and unmapped space (the "frontiers") and keeps going until there's nothing left to discover.

It isn't a separate mode you start. It comes on automatically with 2D SLAM — the moment you call POST /start/slam/2d, the robot starts exploring — and it shuts down when you stop SLAM. (3D SLAM doesn't use it.)

While SLAM is up, you can hand control back and forth. Pause exploration to teleoperate a tricky spot, then resume:

curl -X POST http://<robot>:5000/explore/stop   -H 'Content-Type: application/json' -d '{}'
curl -X POST http://<robot>:5000/explore/resume -H 'Content-Type: application/json' -d '{}'

Check where it's at:

curl http://<robot>:5000/explore/status

The message field is a JSON-encoded string — decode it for running, paused, complete, an info string, and status_received (whether the explorer has reported in yet). Once complete is true, save the map with POST /maps/save before you stop SLAM.

In the OpenMind portal, exploration is part of the SLAM view — you'll see the map fill in automatically, with controls to pause and resume coverage.

If pause and resume seem to do nothing, 2D SLAM probably isn't running — the explorer only exists while it is. See Mapping & SLAM for the full mapping flow, and the Autonomy API Overview for endpoint details.

Last updated

Was this helpful?