Self-driving cars are set to revolutionize the automotive industry, but taking the now nascent technology mainstream will require highly accurate maps. HERE, a Nokia company, believes it has this part figured out.
HERE is mapping major roads at 10-to-20 centimeter accuracy by collecting trillions of 3D points, which are used to model road surfaces down to the number of lanes and their width. “HERE is uniquely positioned to provide the highly accurate and always-fresh 3D maps and cloud services required to enable such an experience,” says Ogi Redzic, Senior Vice President, Connected Driving at HERE. “Precise maps, combined with in-car connectivity, creates endless possibilities.”
Early in 2014 HERE worked on just such an autonomous car solution for the Mercedes S500.
“The technology we developed with Mercedes-Benz allowed us to provide the foundation for automated driving. We took next-generation mapping to its full potential. We collected, processed and published all the data to build an entire new virtual reality of the route the S500 took,” says Redzic.
The S500 has radar sensors and stereoscopic cameras that provide the car with a 360-degree view around the vehicle. These are coupled with intelligent controls, which identify other cars, pedestrians and obstacles. The controls also process the input from the sensors to anticipate what’s going to happen around it. They then assist the driver to avoid accidents by, for example, by stopping the car before it collides with an object.
Commercial fleets will also benefit. HERE and Continental are working on solutions to create Electronic Horizon, an end-to-end connected driving platform for car and truck manufacturers. “With the connected Electronic Horizon vehicle manufacturers will be able to reduce their vehicles’ CO2-footprint by at least two grams per kilometer,” said Ralf Lenninger, Head of Interior Electronics Solutions, Continental in January. “In addition, exact map data will make the driving experience more enjoyable and safe by providing useful information about road congestion so that drivers can predetermine alternate routes as well as providing way in the future to adapt LED headlight functions to road conditions.”
In recognition of these technological advances the BMW Group presented a BMW Supplier Innovation award for Connected Drive to HERE in October 2014. Klaus Draeger, member of the Board of Management of BMW, responsible for Purchasing and Supplier Network, explained: “Innovation is a decisive factor in a vehicle’s market success. Many customers ultimately opt for the car with the most impressive innovations. In fact, innovations are a key driver in our role as a pioneer in shaping the mobility of tomorrow.”
Automotive Industries (AI) asked Redzic to list some of the recent break¬throughs in online map updates made recently by HERE.
Redzic: I think the breakthrough is really in the cloud infrastructure that we are laying down. We are building a powerful location cloud with the aim of being able to deliver map content and services over the air at ever faster rates. The reason we’re making these significant investments in the cloud is that maps are always changing and there is a need to deliver the freshest map data to cars, whether it’s street changes, speed limit changes or real-time traffic information.
AI: What does the BMW award mean for HERE? Tell us about the work you do with the BMW Group.
Redzic: It was a tremendous honor for HERE. We’ve been working with BMW for a long time, providing them with what we believe is the best mapping and navigation content available on the market today. So the award really validated the work that we have done with them. And we’re of course excited to be exploring the future with BMW too.
AI: How important are accurate maps and regular updates for autonomous driving?
Redzic: Our view is that both will be extremely important for the car to be able to drive effectively. The car needs to be able to localize itself. To be able to do that, the car’s sensors need to access a very detailed and accurate map. That’s why we’re building a map with 10-20cm precision, so that the car can figure out which lane it is in and how far away it is from the curb. So, an accurate map is critical.
But the car also needs to know what lies ahead – and by that, I mean what lies beyond the reach of the car’s sensors. And that’s where continuous updates to the map come in. When we talk about continuous updates what we really mean is information continuously flowing to and from the car. And as more and more cars connect to each other and share information, we can build a very precise picture of what is happening on the road. It all makes for a much better automated driving experience.
AI: How feasible will updates of up to 10-20 cm accuracy be?
Redzic: That’s the level of precision we’re mapping to right now and we do that using the most advanced LiDAR technology. Our mapmaking process captures important details such as the slope and curvature of the road, lane markings, the height of bridges, and sign posts, including what that signage denotes. We’re ending up with what we’re calling a HD – or high definition – map.
The idea is that we would map the world with LiDAR completely once and after that would keep the map fresh by taking advantage of sensors from the cars using the HD map. As the number of autonomous cars increases, the sensors on the cars build a view of what they see while driving, which we could cross reference with the HD map. When the two don’t match, the car would send us a signal. When you’ve gotten a few signals identifying the same mismatch, we would update the map ourselves. When there are major structural changes such as new roads, HERE would send out a fleet to map the new changes.
AI: How important is the ‘humanizing’ of driverless cars? What do you mean by it?
Redzic: I referred earlier to the importance of having a highly precise map and being able to keep it updated with real-time information about what is happening on the road. Well, the question you ask is really the third piece of the puzzle. We have to make the driving experience feel like a normal driving experience. We’re also spending a lot of time on this.
If you are in the fast lane of a three-lane highway and decide you’d like to take the next exit ramp, you’d probably move over to the middle lane, have another look, and then move over to the slow lane before indicating again and taking the exit ramp. If you’re in an autonomous car programmed to drive according to the specifications of the road, the whole thing would probably be done in one single maneuver. It could be quite the white knuckle experience.
People need to be able to trust the car to drive for them, and for that trust to happen the car must drive like them instead of a robot. One of the advantages we have is that we’ve collected millions of data points on how people actually drive. Aside from negotiating lane changes, it could be the speed at which people take a curve. The probe data might show that drivers are actually taking a certain curve at 35MPH, even though the speed limit for that section of road is 45MPH. This will go a long way towards making people feel comfortable behind the wheel of a driverless car.
AI: So do you think partially self-driven cars are more likely to be a reality rather than completely driverless cars?
Redzic: Society needs time to adequately adapt to this new automotive era, especially from a legal, regulatory and ethical standpoint. For that reason, the technology is being deployed in increments. Today we already live in a so called “Partially Automated” era, where cars are outfitted with adaptive cruise control and other such features.
In the 2018 to 2020 time frame, we will move towards “Highly Automated Driving,” where the driver will still need to be awake, sober and ready to take control of the vehicle within seconds, should the vehicle require help. By 2025, I expect the legal and regulatory framework will have caught up with the technology, and “Fully Autonomous” vehicles will be introduced. Ultimately it’s about working towards a future where our roads are safe and being in a car is an enjoyable experience.