BERLIN—Thanks to advances in sensor and artificial intelligence, autonomous vehicles are almost as smart as competent human drivers. But while a person instinctively knows how to quickly deal with, say, a road or lane closure due to construction, autonomous vehicles could easily become confused and take too long to deal with the situation as traffic backs up.
This is why we're seeing advances in mapping software, as well as acquisitions and partnerships that focus on the need for hyper-local, constantly updated maps. Part of the motivation for Intel buying sensor-maker Mobileye for $15 billion was to gain access to the company's Road Management Experience, which crowd-sources images from cameras in millions of cars to create highly accurate and constantly updated maps.
Similarly, German automotive supplier Bosch partnered with mapping giant TomTom to develop a platform that captures billions of reflections created by a car's radar signals bouncing off roadside objects. Called Radar Road Signature, "it uses radar sensors to always measure objects so you get a kind of fingerprint of the road," Gerhard Steiger, president of the automotive supplier's Chassis Systems Control Division, told me in an interview last week.
"You can upload that to the cloud and add it to the navigation map and really localize a car," Steiger explained. "Maps will have to be more localized. Regular navigation maps won't cut it since autonomous vehicles need more precise data."
Here, which was acquired two years ago by three German luxury automakers for $2.7 billion to build mapping software, is also working on localized and what it calls "self-healing" maps. (Here is also working with Mobileye and BMW on the REM platform to constantly update its maps.)
In addition to getting data from Audi, BMW, and Mercedes-Benz vehicles, Here also provides mapping software for four out of five cars in North America and Europe with in-dash navigation. So it's uniquely positioned to supply next-generation maps for autonomous driving and is creating what it calls a Reality Index for self-driving cars to better grasp their surroundings.
Rich Data From Premium Cars
As with Mobileye's REM and the Bosch-TomTom Radar Road Signature, Here is collecting data from vehicles using sensors and camera. But in this case, "it's very rich data since from premium cars made by our shareholders," Ralf Herrtwich, head of automotive, told me in an interview this week at the company's HQ in Berlin.
"When we build our HD Live Maps, there's two portions to it," Herrtwich explained. "The original collection of scanned roads. But then roads change—even if it's just a construction zone—and it's necessary as soon as something changes to have that reflected in the map.
- With Driverless Tech, It's Important to Sweat the Small Stuff With Driverless Tech, It's Important to Sweat the Small Stuff
"A vehicle that detects something divergent on the road sends us that report," Herrtwich noted. "We may not trust every report, but if three vehicles report the same thing, you know something changed and you adjust the map according to what we learn—and then we have a self-healing map."
Here is also looking beyond cars and collecting data for its Reality Index from drones and IoT devices to provide even more granular location data for autonomous vehicles and future mobility services. "It will be a database that provides information in real time to self-driving cars about everything that's happening in the environment," Herrtwich said.
Then autonomous vehicles may be more aware of what's around them than humans.