Autonomous driving rain and low-light LiDAR-camera
On-road or closed-course AV capture emphasizing weather and lighting tails, with intervention and disengagement metadata.
This is a custom capture program for autonomous driving in rain and low light, the conditions where perception degrades most. We record synchronized LiDAR and camera segments on road or closed course, tagged with intervention and disengagement metadata. Clear-day data overstates perception reliability; these weather and lighting tails are what robust AV evaluation actually needs.
What we collect
Weather-tail segments with intervention and disengagement metadata for perception robustness evals, spanning rain intensity, glare, dusk, and night conditions.
Sensors and modalities
LiDAR and camera matched to your production rig, time-aligned through our multi-sensor synchronization service so fused perception is trustworthy under degraded conditions.
How capture works
A pilot confirms sensor parity, authorization, and labeling, then capture scales across weather windows and routes following our robotics data collection workflow with safety drivers.
QA and metadata
Episodes carry weather, lighting, geolocation, and intervention/disengagement tags. QA covers sensor sync, calibration, and label completeness against agreed criteria.
Who it is for
Autonomous vehicle teams hardening perception against adverse conditions. The tail focus relies on deliberate edge-case data collection for rare interventions.
Scenario FAQ
Permits, consent, and safety drivers per jurisdiction are defined before collection; closed-course capture is available where on-road is impractical.
Rain, glare, and low light are where perception degrades most. Capturing these tails deliberately is how evaluation reflects adverse conditions.
Yes. Sensor parity with your production vehicle is a core scoping requirement so the data transfers to your stack.
Scope your capture program
Book a discovery call to align on your stack and data requirements.