Skip to main content
The final leg of the delivery journey is the most expensive part, and often the least well mapped. While longstanding road networks are well covered, the sidewalks, crosswalks, and building approaches a courier or a delivery robot actually uses are usually missing. Wherobots can build that missing layer from imagery, join it to address and building data, and model the service areas that your dispatch system plans against.

​
How to use these solutions

Treat these notebooks as data engineering patterns as you evaluate Wherobots for your business or organization:

Pedestrian Network Mapping

Detect sidewalks, crosswalks, and pedestrian pathways from aerial imagery and turn them into routable geometry for couriers and sidewalk robots.

Delivery Zones & Service Areas

Generate walk- and drive-time catchments around depots, dark stores, and lockers to size delivery zones and set realistic promise windows.

Address & Building Context

Resolve drop-off points against building footprints and points of interest so a stop lands at the right entrance, not the street centroid.

Road Coverage in Under-Mapped Areas

Detect roads from imagery where commercial basemaps are thin, so rural and newly built routes are plannable.

​
Key Capabilities

Run RasterFlow’s Tile2Net model over aerial imagery to detect sidewalks, crosswalks, and pedestrian paths across a whole service area. Vectorize the output into routable geometry, then inspect it for gaps and discontinuities to judge which segments carry a delivery robot or a courier on foot.

Detecting sidewalks with RasterFlow

Turn detected pathways into vector geometry you can route on.
Score a retail corridor or neighborhood on the continuity of its pedestrian network, then join that score to demand and store locations to decide where walking beats driving for the final leg of the delivery journey.

Detecting sidewalks with RasterFlow

Map sidewalks and crosswalks, then score the corridor for continuity.
Generate isochrones from every depot, dark store, and locker to size zones by travel time rather than radius. Recompute them as the network changes to keep promise windows honest.

Making drive time isochrones

Size delivery zones by travel time rather than radius.
Join delivery addresses to building footprints and points of interest so stops resolve to real entrances, and cluster historical orders to find where density justifies a locker or a micro-hub.

Using Overture Maps

Resolve stops against building footprints and places.

DBSCAN clustering

Find where order density justifies a locker or micro-hub.
Apply RasterFlow road detection to imagery in rural and newly developed areas where commercial basemaps lag, so routes exist for addresses the map does not know about yet.

Detecting roads with RasterFlow

Find roads in areas where commercial basemaps are thin.
Generate vector tiles and PMTiles directly from WherobotsDB so ops teams, partners, and customers can view live coverage, zone boundaries, and pathway networks in any web map, with no tile server to operate.

How to generate PMTiles

Publish a single-file tile archive of your coverage.

Creating vector tiles with VTiles

Render zones and pathways in any web map, no tile server.

​
Wherobots Advantage

Get started with Wherobots

Try Wherobots with your own geospatial data in minutes.