Sidewalk last-mile delivery capture
Sidewalk and campus last-mile delivery navigation with RGB or RGB-D, pedestrian encounters, and curb-event metadata.
This is a custom capture program for sidewalk and campus last-mile delivery navigation. Operators or supervised autonomy run your robot along real paths while we record cameras, pose, and encounter labels for pedestrians, curbs, and blockages. It exists because sidewalk right-of-way is not a warehouse aisle and not an on-road lane, and those two existing corpora miss the yield and curb events that stop a delivery robot.
What we collect
Route episodes with pedestrian encounters, curb crossings, and blockage outcomes across lighting and traffic density. Yields, aborted crossings, and forced detours are labeled so evaluation can slice a clean run from an intervention.
How capture works
Scoping locks route class, sensor suite, and consent rules. A pilot checks logging and privacy redaction, then capture scales across shifts using robotics data collection. Crowding and blocked paths follow edge-case data collection.
Who it is for
Delivery robotics teams whose robot must share sidewalks and campuses. This scenario belongs to delivery robotics data collection. The manipulation sibling is curb package handoff.
Scenario FAQ
Scope your capture program
Book a discovery call to align on your stack and data requirements.