One engine, inside or outside the vehicle.

Same build, same measurement, same evidence. Three places it runs, one picture it watches.

Off the vehicle, on recordings you already have.

A year of drive logs, played back. The engine goes through them sensor by sensor and finds the one that drifted while every recording looked fine.

On the vehicle, the same build as a monitor.

Same picture, seen live. When a sensor leaves its own baseline it raises an event, with severity, to the function supervisor. Advisory only, it never actuates.

Across the fleet, both feed one view.

Which sensors are degrading, why, and when they hit their limit, weeks ahead. On your infrastructure, and each OEM sees only its own vehicles.
Front camera, night streetthe picture every place below watches

A pedestrian crossing ahead, a car stopped at the line. Clean, on day one.

Recorded drive logsone sensor, one year, played back
JanAprJulOctDec front camera, watch, first event warn, work order

recordings analysed 0

Signal integrity, liveper sensor, against its own baseline
front camerapass
sharpness at baseline
rear radarpass
signal to noise 99 %
roof lidarpass
valid returns 100 %
IMUpass
bias at baseline

event front camera, watch, sharpness −10 % · to function supervisor, advisory

Fleet, ranked by time to limityour vehicles only
VehicleSensorCauseLimit in
2183front cameradefocus3 weeks
0912rear radaroff axis5 weeks
3350front camerasoiling9 weeks
front camera onboard self-testOK obsurverpass, sharpness at baseline

How it fits together.

Two inputs, one engine, one output. On the vehicle the result also goes back to the function supervisor, as an advisory.

Recorded drive logs the recordings you already store JanDec no onboard integration no code in your pipeline where every project starts The vehicle, live on the vehicle compute, as a monitor now no cloud connection needed obsurver engine one build, both places Measures every sensor against its own baseline. camera, radar, lidar, IMU, GNSS, any supplier own baseline limit today front camera sharpness −18 % vs baseline watch Advisory result integrity per sensor, a degradation event with severity, and the evidence record SOTIF and type approval ask for. On the vehicle, the event also goes back to your function supervisor. Advisory only, it never actuates. One fleet view fed by both Which sensors are degrading, why, and when they reach their limit. vehiclesensorcauselimit in 2183front cameradefocus3 weeks 0912rear radaroff axis5 weeks 3350front camerasoiling9 weeks Weeks ahead, so parts and workshop time get planned, not discovered. Each OEM sees only its own vehicles. illustrative values Recorded drive logs the recordings you already store JanDec no onboard integration no code in your pipeline where every project starts The vehicle, live on the vehicle compute, as a monitor now no cloud connection needed On the vehicle, the event also goes back to your function supervisor. Advisory only, it never actuates. obsurver engine one build, both places Measures every sensor against its own baseline. camera, radar, lidar, IMU, GNSS, any supplier own baseline limit today front camera sharpness −18 % watch Advisory result integrity per sensor, a degradation event with severity, and the evidence record SOTIF and type approval ask for. One fleet view fed by both Which sensors are degrading, why, and when they reach their limit. vehiclesensorcauselimit in 2183front cameradefocus3 weeks 0912rear radaroff axis5 weeks 3350front camerasoiling9 weeks Weeks ahead, so parts and workshop time get planned, not discovered. Each OEM sees only its own vehicles. illustrative values

Off the vehicle

Where every project starts. Offline, over the recordings you already store.

in
synchronized sensor frames and object lists, from any recorder
out
integrity per sensor, degradation events with severity, the evidence record
touches
nothing in your stack

On the vehicle

The same build on the vehicle compute, once it has proven itself on your data.

reads
the pre-perception and post-perception interfaces
writes
nothing into the vehicle
reports to
your function supervisor, advisory only

Across the fleet

One view fed by both. Runs on your infrastructure.

shows
which sensors are degrading, why, and when they reach their limit
plans
parts, workshop capacity and claims, weeks ahead
scope
each OEM sees its own vehicles only

Start on a drive log, not an integration.

30 minutes with the founders. Bring a recording if you have one, we bring the software.