AI that writes robot code and proves it works.

Describe what you want your robot to do. Quatern writes the navigation, mapping, and localization code, tests it on your robot's real sensor data, and watches it run.

Try it free

$ pipx install quatern
$ quatern
quatern · turtlebot4
>
reading the robot config and the lidar's /scan topic
✓ wrote lidar_stop.py
+ def on_scan(scan): + ahead = scan.ranges(-15, 15) + if min(ahead) < STOP_DISTANCE: + robot.stop()
✓ verified on the recorded drive
stops 0.5 m before obstacles
◆ proceed past the gate? the robot will move [y/N]
running, watchdog on track
✓ run recorded

Works with

  • ™ROS 2
  • Python
  • C++
  • TurtleBot
  • Any robot with a URDF

How it works

  1. Step 1

    Writes it

    Describe what you want in plain English. The agent writes navigation, mapping, and localization for your robot, in any language.

  2. Step 2

    Debugs it

    Hand it a failing launch. It reads your logs, TF tree, and URDF, finds the root cause, and writes the fix.

  3. Step 3

    Proves it

    Every change is replayed on your robot's recordings, with before and after metrics, before anything moves.

  4. Step 4

    Watches it

    A live watchdog follows the robot while it runs, and every run leaves a record of what happened.

You stay in control: every change is a diff you can read, tested on your data, and nothing moves without your approval.

Stops the robot the moment reality drifts from the plan.

A real Quatern run in the simulator. Inject drift and watch the watchdog stop it.

  • Nothing deploys to a real robot without a person approving it
  • Stops on drift, a sensor that goes quiet, or a sensor that disagrees with the others

turtlebot4 · hallway world · replayed from the run record

Have a robot? Let's get it running.

Tell us about your robot, and we'll get it running with you.

Try it free

$ pipx install quatern
$ quatern