
To acquire our data inputs, we purchase high-quality aerial imagery, which costs 8-10 times less than lidar data inputs.
To acquire our data inputs, we purchase high-quality aerial imagery, which costs 8-10 times less than lidar data inputs.
We process the aerial images through our one-of-a-kind computer vision technology. The AI recognizes and extracts features (like road edges, lane markers, signage, etc.) from the built environment. Unlike with the lidar mapping process, the data doesn't require filtering, alignment, or classification, which substantially lowers computational costs.
The resulting feature map provides a geometric representation of the area that is accurate within 10cm.
The map undergoes a manual quality assurance process to ensure accuracy.
Our AI annotates the map with semantic information, like directionality, lane center lines, and intersection curvature. This is critical for autonomous vehicle operations.
The map undergoes another manual quality assurance process.
The HD map is ready! The total process, from data collection to the final product, takes 7-10 days for a mid-sized city.
Our maps are available in all industry standard formats, or they can be custom-built to your platform.