Post-processed yesterday's site test
3 relevant patterns detected across 14 flights.
- Flight logs14 MCAP files
- TelemetryPosition, IMU, setpoints
- Robot codeRelease 4.3, 12 commits
Your team says what the robot should do and which limits it must respect. We're building the flywheel that does the rest, starting with a shadow engineer that takes you from field failure to validated redeploy.
Your shadow
engineer
So your team focuses on what's new, and what you fixed stays fixed.
3 relevant patterns detected across 14 flights.
>
nonlimi · get_incident NL-2481
Jump of 1.1 m at handover, in 3 of 14 flights
nonlimi · get_analysis NL-2481
Reproduced at 96%. The planner keeps its old goal after teleop.
Read src/planner/local_planner.cpp
Update src/planner/local_planner.cpp
- keep_goal(state);
+ reset_goal_on_handover(state);
nonlimi · run_scenario teleop_handover
24 of 24 variations pass. Largest jump 0.3 m.
Fixed. Sent 4.3.1 to the release suite.
Every site is different and field days are scarce. Compute isn't. So the loop keeps going after your team goes home, and a better release is waiting in the morning.
Logs, ROS bags, config, code, site videos, tickets. It's a messy world, and nobody has time to clean it up. Nonlimi makes sense of it as it is.
| Scenario | Pass when |
|---|---|
| Pilot hands back to autopilot in a gust | no jump above 0.5 m |
| Lidar drops out on the approach | lands in a safe zone |
| GNSS lost under the bridge, 20 s | drift under 2 m |
| A person walks under the flight path | never over a person |
| A second drone crosses the corridor | keeps 5 m apart |
| + 4 more, each with its evaluators | |
Writing scenarios by hand is the slow part, and the worst cases are the ones nobody thought of. Agents propose them, each with the evaluators that decide pass or fail.
Every release leaves scenarios and results behind, and the next run starts from them. Nonlimi re-runs only what a change touches and tells you which earlier results still stand.
Every build or parameter change runs your full suite in parallel, in a simulation kept true to your field data. You see what regressed and why, and get a report to sign before anyone drives to the field.
| Where it could come from | Evidence | |
|---|---|---|
| Camera calibration | reprojection error normal | ruled out |
| Localization | drift under 1 mm | ruled out |
| Mechanical flex | boom bends 9 mm under load | root cause |
| Control timing | no missed cycles | ruled out |
| Planning logic | a patch would hide it | not the cause |
| Site model | scan matches the wall | ruled out |
A misplaced brick could come from camera calibration, localization drift, mechanical flex, control timing, planning logic or an incorrect site model. An agent may propose a convincing software fix while the underlying cause remains.
It is designed to compound: each release leaves fewer problems for your customers to find, and teams can release more often and at higher quality.
Field tests go to what you are building now. We agree these four numbers with each team before we start.
Example: columns are releases, rows are scenarios
If you ship robot software, we'd like to hear from you.
Talk to us