The Japanese carmaker will debut a map-generating system at CES next month. The system uses data from cameras and GPS units installed in production cars to build accurate maps. Images and video are harvested from cars as they drive along, and sent to data centers to be pieced together into a picture of the terrain.
Toyota claims using its own production cars to gather data will produce more accurate maps than using dedicated vehicles, like the sensor-festooned cars used for Google Maps. It argues that using cars already on the road will allow for real-time updating of maps, and will be more cost effective than dedicated data-gathering runs. Granted, it’s unclear how many Toyota drivers will volunteer to turn their cars into mobile data mines.
While Toyota admits that a system based on cameras and GPS has a higher probability of error than the laser scanners used by existing systems, it claims to have solved that problem. Any potential errors will be mitigated by technologies that integrate data from multiple vehicles and “high-precision trajectory estimation technologies.” Toyota claims its system is accurate to within 5 centimeters on straight roads.
Deployment of the map-generating system will initially be limited to highways, Toyota eventually plans to use it on other types of roads, and to “assist in hazard avoidance.” The carmaker believes this technology will be important in autonomous-driving systems it plans to offer on production cars beginning in 2020, and eventual fully autonomous vehicles.
In a fairly short period of time, Toyota went from being somewhat disinterested in self-driving cars to launching an ambitious plan. The carmaker is collaborating with MIT and Stanford to develop artificial intelligence for future self-driving cars. But rather than replace human drivers, Toyota believes this tech can complement them, and encourage a relationship between driver and car.
