Right here Applied sciences has efficiently fended off competitors from know-how giants seeking to transfer into high-definition mapping, which is likely one of the key parts for autonomous driving. CEO Edzard Overbeek lately advised Automotive Information Europe Correspondent Nick Gibbs how the situation knowledge and know-how platform supplier will maintain a step forward of Google.
Volvo and Google used CES to announce a partnership on HD maps. How do you battle the tech giants?
We do fairly properly. For instance, the Mercedes S-Class with Degree 3 autonomous functionality makes use of our high-definition maps. We simply launched with the BMW i7, which is at the moment Degree 2 Plus and can go to Degree 3. So far as I do know, we’re the most important provider of HD maps in industrial contracts. Google has a relationship with Volvo, however we even have a relationship with Volvo, not for HD maps, however on different issues.
What are your strengths towards somebody like Google?
They do not have our capabilities. For instance, in Europe we’ve got 85 p.c of the marketplace for the brand new ISA [intelligent speed assist] regulation, which must know the pace restrict of the street you are driving on [to provide feedback to the driver if the posted speed limit is being exceeded]. The tech corporations do not have that. The second is privateness. Firms are beginning to notice that if you use Google Maps, you might be principally promoting your knowledge to Google for them to construct higher ads. We do not do this. The third is that we’ve got no worry of competitors. We’re impartial Switzerland relating to location knowledge. We’re not doing the rest with it. In the event you do enterprise with the Googles of the world, then they could come into your trade.
The place are your largest threats coming from by way of startups?
It is a very related query, however I might in all probability have answered it in a different way 5 years in the past when all people and their moms have been moving into mapping. At present, only a few are left. Deepmap has been acquired [by chip provider Nvidia]. Civil Maps simply acquired acquired [by lidar company Luminar]. So, of all of the startups round mapping, I believe only a few have survived.
They have been purchased however they nonetheless exist, so aren’t they a danger?
They acquired absorbed right into a stack, changing into a part of a vertically built-in system. That’s not actually a mapping competitor for us. Individuals are beginning to notice the significance of location knowledge for all use instances, not simply the autonomous automobile. Nonetheless, constructing that is very costly and really advanced. We’ve got seen individuals strive. And, sure, it is nice to place a lidar on a automotive, drive Interstate 280 [a major north–south highway in the San Francisco Bay Area] after which say, “I’ve acquired a high-definition map of I-280.” That is advantageous, however now do this for 200 international locations around the globe in close to actual time with refresh charges of 24 hours or much less in each decision kind.
Are you able to collect knowledge from buyer automobiles like Google will do from the brand new Volvo EX90 SUV?
Sure, we’ve got 35 million automobiles offering us sensor knowledge that we make maps from. We extract the view from cameras and make a digital illustration. That’s what we do at scale. In our new Unimap we’ve got 500 million kilometers of roads being processed each hour. So, it is nice that Google has now entry to Volvo and I do not wish to downplay that, however it’s a bit of bit completely different.
How are you downloading that info from automobiles?
We do not do the complete picture. It is not the complete two terabytes of knowledge as a result of that will be uneconomical. Producers permit us entry from knowledge feeds from the automotive sensor cluster.
What do the producers get out of you extracting their knowledge?
They get higher maps. They get location-based providers. We’re sitting down with them on a complete array of recent experiences relating to navigation. So, what’s the subsequent technology navigation seem like? Is that solely within the center console or are we going to do OLED see-through shows and different shows within the automotive?
Are you able to extract high-definition knowledge from buyer automobiles?
Sure. You want lidar, which has millimeter precision. Radar and ultrasound should not excessive definition. They might be sooner or later however not in the present day. We’ve got arguably the most important lidar database on this planet.
You introduced at CES one other technique to ship merchandise, Unimap. How does that work?
Previously we began with one pipeline — customary definition maps. Then we construct one other pipeline that offered visitors knowledge. These are principally automated flows from a database that steers into the system — a automotive, a cellphone, a drone, a robotic … no matter. Then we added locations knowledge as a brand new pipeline, then high-definition knowledge, then 3D maps. What we’ve got accomplished with Unimap is combine all of them into one pipeline.
What’s the benefit?
Now we align the info feed so that they make logical sense to one another. 5 pipelines into one. No person else has this within the trade. For instance, the visitors gentle info was not linked to the street info. Now it’s. So, the street segments turns into a lot richer.
How does this profit automakers?
They are saying that is what they want. With 5 completely different feeds, it turns into a spaghetti and with a really excessive value to combine. Having one pipeline is way extra value environment friendly.
What occurs if somebody simply desires one aspect of that pipeline?
You simply extract, for instance, customary definition maps or simply street curvature or tunnels. You solely pay for that. It turns into a way more modular technique to devour map knowledge.
Is that this all downloaded into the automotive?
It’s in the present day however sooner or later you are able to do that by way of the cloud.
What’s This is automotive/non-automotive cut up?
About 70 p.c automotive and 30 p.c non-automotive. The technique is to develop each. We’ve got a pretty big share of automakers already, however we are attempting to broaden that. In non-automotive, our focus is on transport and logistics, retail and e-commerce.
What’s your geographic cut up?
In all probability 40 p.c U.S., 40 p.c Europe, 20 p.c Asia. China is a vital market, however we do not function straight in China. We’ve got a three way partnership with NavInfo that serves the inbound [non-Chinese] automakers there.
The place is the following large development in automotive?
We’ve got a lion’s share in automotive. We’ve got two new merchandise, one is street alerts, for instance, if the automotive forward has damaged down, we will show that. We even have autonomous driving zones, which is principally geofencing the map. If there may be an surroundings the place you do not need the autonomous driving system to be energetic, you can also make positive it could’t be activated there. Or, if there are specific occasions the place you do not need the automated driving system to be energetic, you possibly can principally shut it down.
What does a high-definition map imply?
Maps are for human beings. A automotive wants an instruction set; it wants machine language. That’s the place we’re shifting the corporate, towards extra of a location, big-data firm. A machine simply takes the feed of a lane or a street curvature or a bridge or no matter.
Some corporations have stated that hands-free driving might be attainable with out excessive definition. Are you seeing that development?
During the last couple of years, sure corporations have stated sensors within the automotive are ok, however I believe all people has stepped away from that thought. The belief now could be that in all probability crucial sensor within the automotive is the map, and that every little thing feeds into that. A map seems across the corners, proper? So, in automated driving, the automotive turns into conscious of the state of affairs and appears by buildings and round corners.
Is there something automakers can take from the high-definition map and switch it into one thing fascinating for the driving force?
Sure, from a location-based providers standpoint. As an example, take into consideration a constructing and downloading the historical past of that constructing onto your automotive window. We’re pondering alongside these strains.
What does having automakers as shareholders do for you? Does it shield you? How do they profit?
It is very strategic for them. They took the choice to purchase the corporate as a result of they consider that the aggregation of knowledge is important for the automotive trade normally, not only for the shareholders. It is about ensuring that they’re impartial from the large know-how corporations. I have a look at them as shareholders. We’ve got very completely different conversations with them as shareholders than we do after we are speaking about industrial contracts with them, and we do not combine the 2.
Do your shareholders generally do enterprise together with your rivals?
Sure. There isn’t any assure that they provide us the enterprise simply because they’re shareholders. We have to carry out. That’s the approach it is arrange. For instance, BMW has a take care of Mapbox, and Nvidia is a accomplice with Mercedes on high-definition mapping. It is my job to be sure that we’re higher. It retains me on my toes.