- What is this and how does this work?
- Why is it important?
- Where are you crowdsourcing map data?
- How good will this crowdsourced map data be?
i. What is this and how does this work?
Edge Mapping, which occurs in-vehicle and utilizes onboard hardware and software, is Civil Maps’ process of creating, maintaining, and using high definition, machine-readable maps while driving. To enable this, we’ve taken some traditional SLAM (Simultaneous Localization and Mapping) techniques widely used in robotics and updated them to facilitate global scale map building, consumption, and crowdsourcing. Edge Mapping™ is similar to “edge computing” or “fog computing,” wherein data is processed near the source. In our case, after raw sensor data is processed in-car, we send it to our cloud infrastructure over cellular networks (existing 3G or 4G works). From there, we organize and aggregate the data generated during the initial passes performed in-vehicle and we then provide the resulting map info back to other cars in our network. Developers can quickly get started edge mapping with our Atlas DevKit. They can also easily use their existing sensor configurations via our hardware abstraction layer (HAL).
Back to Top
ii. Why is it important?
The initial phases of autonomous driving include creating and updating maps in small geofenced locations. However, the addressable market for autonomous cars is closely tied to the areas that are mapped and continuously updated. Without access to areas with machine-readable maps, autonomous vehicles are sub-optimally operable. In order to cover larger areas and to deploy at scale, the capital expenses involved need to be minimized. Traditional mapping techniques involving large fleets of dedicated mapping vehicles and workers are extremely costly. Many companies building and updating high definition maps for autonomous cars are spending billions yearly just to map geofenced areas. Expanding beyond that with the conventional approach could take significant time and will be cost prohibitive. In our opinion, crowdsourced edge mapping is the best economic solution to building a base of high-quality maps and keeping them updated.
Back to Top
iii. Where are you crowdsourcing map data?
Civil Maps is edge mapping with a handful of OEMs, our own development vehicles, and other fleets of cars, both “in the wild” and in controlled geofenced areas, internationally.
Back to Top
iv. How good will this crowdsourced map data be?
Not every map contribution needs to be perfect. As we gather more map data, we are able to aggregate, validate, and extract the correct information on the server side. By leveraging a crowdsourced approach, Civil Maps’ network can get minute by minute updates as opposed to a map that is updated every 3 to 6 months, a typical timeframe with conventional mapping methods.