refactor(control-station): separate lidar diagnostics from datasets
This commit is contained in:
@@ -69,6 +69,12 @@ unavailable rather than inventing timestamps.
|
||||
## Consequences
|
||||
|
||||
- The UI must label accumulated K1 evidence as a map product, not a scan.
|
||||
- Real-sensor diagnostics live under Fleet and never mix public dataset frames
|
||||
into device evidence.
|
||||
- Dataset catalog, preview and qualification-run entry live under Polygon;
|
||||
Data remains the source-of-record and artifact store.
|
||||
- Until bytes are installed, the UI exposes an explicit empty state rather
|
||||
than presenting this architecture contract as an interactive tool.
|
||||
- Dataset and device inputs can share downstream algorithms only after their
|
||||
normalized contracts match.
|
||||
- Sensor adaptation may change range, FOV, point density, noise and dropout for
|
||||
|
||||
Reference in New Issue
Block a user