STRATEGIES Interview Dr. Martin Birkner International Transportation (67) 2 | 2015 10 “Highly automated vehicles are not built for their own sake.” Self-driving cars are certainly in vogue. They are one of those technological developments that analysts reckon will have great market potential across the world. But what do vehicles actually need to make their independent way through the streets of the world? How far have we come? And how quickly are we going to have to get used to swarms of “robot cars”? These are some of the questions International Transportation asked Dr. Martin Birkner, Marketing Manager Automotive Sector at Here, Nokia’s mapping service. Dr. Birkner, some users know the name Nokia Here from the mapping service provided on their Windows smartphones. But after the announced takeover by Audi, BMW and Daimler, automobile manufacturers suddenly seem to be showing serious interest in your products and services. What’s behind it all? For a start, we have been working closely with the automotive industry for decades, right across the world. The automotive industry has confidence in us, and we continue to make a contribution to our customers’ success through innovation, quality improvements and product enhancements. In Europe and the USA, you will find our- maps in four out of five cars that have built-in navigation systems, and as far as traffic information goes, we have the world’s largest coverage with Here Real Time Traffic, our traffic information service that-offers 100 % road coverage in 50 countries. The reason for the even greater interest in our business is the fact that the ‘maps’ that we use in our navigation system these days are currently going through a fundamental transformation. As part of the trend towards digitization and connectivity, these maps are being converted into something that we call the ‘Here Location Cloud’. In the future, this cloud will be a central component for the vehicle connectivity and for all of the underlying business models for infotainment and navigation in the context of highly automated driving or even autonomous driving. The Here Location Cloud consists of three components and provides the functions that will be essential for the automotive companies to strengthen their competitive position. What components are these? The first component includes what we have been doing since Nokia took over Navteq in 2008. We measure the world. At that time the focus was on the creation of 2D maps, whereas today we are working on 3D and developing high-resolution maps, so-called HD maps, which can be accurate to within ten centimeters. A major contribution to the development of these HD maps has been our fleet of around 200 vehicles, which are constantly travelling the world’s roads, capturing highly accurate mapping data. Highresolution mapping - effectively indexing the world - stimulates the imagination, of course. Suddenly we understand exactly where objects such as road markings, bridges or road exits are located, and can pass this information on to the equipment. This is of central importance for highly automated driving. We are already working on projects in this area with leading partners in the automobile industry. In other words, with nothing more than your maps you would be able to tell a car if it is correctly positioned in its lane? That’s right. Lane accuracy is one of the central elements of our HD maps. In conjunction with the vehicle’s technical systems, we can tell exactly where the vehicle is positioned on the highway - for example, whether in the left lane or in an exit lane and help fine-tune the driving strategy of a highly automated vehicle so that it is safe, efficient and enjoyable. In the medium term this will be essential for highly automated driving, but it is also useful today for what we call ‘Smart Guidance’. Current GPS navigation devices are sometimes not sufficiently accurate to precisely display the traffic as it actually is. Based on some recent innovations in traffic guidance data provision (‘Split Lane Traffic’) and our HD maps, we will be one of the first companies on the market to bring this accuracy to dynamic services, such as traffic guidance, and make it available to our customers. For this purpose, the second component is extremely important: the so-called ‘dynamic data layer’. It is not enough to know exactly where things are. We also need to know precisely what those things and people are actually doing and how they move and change. Every movement and every incident - such as an accident - creates dynamic data, which in turn influences other road users. Across the world, millions of road layouts are permanently changing, and we must therefore constantly adapt our maps, so that we can tell the car that it is located on the turning lane to the right, and next to it is a fire hydrant. But of course there are even more important pieces of information, for instance that traffic has come to a halt 50 m ahead, requiring the car to immediately adjust its speed. All this is the so-called dynamic data layer. The dynamic data layer becomes progressively more effective when more data is provided by additional vehicles and other sources. We have therefore recently recommended to the car industry to define and introduce a standard interface format so as to provide sensor data from vehicles on maps in a uniform manner. This would enable us to further improve the foundations for the dynamic data layers. Can this not already be achieved using radar sensors? Sure, in theory, a car could do that today, if the conditions are right. There have been pilot tests, such as cars driving across the US, from the West Coast to the East Coast, using data from the vehicle’s own sensor system. However, these are test vehicles - for the wider market that will not work, for a variety of reasons. Highly automated driving must not remain an application that is limited to just a few high-end vehicles. We must all implement this technology on a global scale so as to fully tap the benefits of highly automated driving, as there are: Interview Dr. Martin Birkner STRATEGIES International Transportation (67) 2 | 2015 11 fewer vehicles on the roads, more efficient road transport, less pollution and a better driving experience for people travelling from A to B. This can only be achieved if we combine a vehicle’s sensor system and computing capabilities with the dynamic services offered by the cloud. Dynamic services such as predictive traffic information can then be used to inform the vehicle - or the driver and the passengers - as early as possible of any decisions that need to be taken. This will help to reduce the stress that we sometimes experience in today’s traffic, and pave the way for highly automated driving. Don’t you expect any difficulties to arise between connected and non-connected driving during the transitional period? To be honest, no, I don’t see any major difficulties, because this is an issue that is already playing out on our roads, albeit in a restricted form. In a sense, connected vehicles have been around for some time already. Their drivers have access to dynamic traffic information, such as constantly updated congestion information displayed on their navigation devices, which enables them to adjust their driving behavior at an early stage. At the same time, other drivers respectively their vehicles are not connected in any way. We are already able to see that with about 15 to 20 % of vehicles in a given geographical transport area being connected, traffic becomes much more efficient, less stressful and more environmentally friendly for all involved. The question is more of an issue in really chaotic urban transport situations. In fact, we are not just talking about the highways in Germany or Switzerland here, but also about mega-cities in China, South America, Africa or India, where completely different traffic conditions prevail. In such places, the transitional phase could possibly be eased by the city council or the government allocating only specific routes for highly automated driving, ideally in cooperation with the automotive industry. Perhaps this could be done temporarily and for individual road stretches only, depending on the sectionspecific risk level assessed on the basis of traffic volume, time of day or indeed the current weather. How do you see that working out? Using our map data in combination with our dynamic services that provide information about the current traffic situation, the next 20, 40 or 100 km on a particular section of road, such as a motorway, can be analyzed. Then dedicated algorithms can be used to calculate the risk. Depending on the result, permission for highly automated driving can be granted to connected, highly automated vehicles over a defined stretch of road. So here we are talking also about data coming from the road infrastructure? Absolutely. We go for cooperation with all the major players in the field of Intelligent Transportation Systems (ITS) in order to incorporate data from traffic lights etc. Highly precise map data, information provided by the infrastructure, data from the vehicle, as well as insights about driving behavior - the combination of all types of data is what makes the Location Cloud so useful. By aggregating and analyzing all this data, we know exactly what’s out there, we also know how the situation is changing dynamically at this very moment, and we can draw conclusions about how the situation is likely to develop in the next half hour. Now the fascinating question is this: What happens with all this information? Here we’re talking about the third component, the Location Cloud. This component makes it possible to offer all of these functions and data in an appropriate form to third parties who either want to use them for business - for example to sell cars or handsets - or who, as end users, need the data and functionality for driving purposes. For these uses, we make the various delivery capabilities of our platform available. Access to this combination of high-resolution maps plus dynamic adjustments is possible either via an Application Programmer Interface (API) or a Service Development Kit (SDK), enabling companies such as automakers to integrate all this data quickly and easily in their own products and innovations in a suitable form. This helps shorten the time to market and dramatically boosts our customers’ capacity for innovation. Until now, innovation cycles usually lasted three to five years. Now they are at under three years - if, in view of the current speed of digitization, we can still talk about innovation cycles at-all. Moreover, with Here Auto we offer a complete end-to-end navigation solution for the automotive industry, which makes our Location Cloud not only available for use in the vehicle, but also on a smartphone. This is one of the first solutions on the market that accompanies the drivers on their entire journey from A to B - not merely in the car, but also when outside the vehicle. For example, Jaguar recently featured Here- Auto in the new Jaguar XF launched this year, where it forms the basis for their navigation solution. This allows the total journey to be planned on a smartphone even before getting into the car, accompanying the driver with useful information during the drive itself and then again on the smartphone over the final phase of the trip. And how does this work? In the medium term, for example, the smartphone can advise the user to start half an hour earlier in order to avoid a traffic jam - or better still, set off half an hour later, because then, according to the data, the congestion will have dispersed. If you think this through and place this scenario in a net- Dr. Martin Birkner has over 17 years of experience in strategy and product development in the areas of connectivity solutions and applications in the automotive industry, smart grids and telecommunications. For more than eight years he was a consultant in the telecommunications industry, and later worked for Robert Bosch GmbH, Telefónica, and frog design. Today Dr. Birkner is responsible for global product marketing in the automotive sector at Here. STRATEGIES Interview Dr. Martin Birkner International Transportation (67) 2 | 2015 12 worked, urban environment, you suddenly discover a range of possibilities for optimizing the entire traffic flow using such applications. This does not only benefit the end users, who will reach their destinations faster, more efficiently and in greater comfort, but the cities as a whole as well. What you have just said about the opportunities such systems offer to the user inevitably leads to the issue of security. In practice, privacy no longer exists when, from dawn to dusk, virtually all our actions, movements and plans are stored in the cloud. And aside from that, any electronic system can be hacked. It will probably never be possible to offer one hundred percent security, but when it comes to our systems, you must distinguish between anonymous and personalized data. The connection of the driver, the vehicle and the passengers is carried out in completely anonymous form. The data is stored in our OEM customers’ protected, highly secure data bases, and none of the data transactions and analyses carried out within the cloud to provide the desired services are based on personalized data. Decentralization is also one of the key principles that will govern the cloud in the future. The ‘Silicon Valley’ approach of using centralized and personalized data in a cloud in the context of the “Attention Economy” to foster sales in the automotive industry is completely contrary to the way that we see things. We see ourselves more as a ‘steward’ of the data, and agree business models under license that do not require any further monetization of the data to become profitable. Another principle is the protection of data integrity. Just think of the various events that have occurred recently, where vehicles were taken over and controlled remotely from the outside: Such nightmare scenarios, where the driver has no way to stop the vehicle or to get out, are precisely the reason why the automotive industry needs to create its own cloud aimed at securely protecting all those sensitive data-… … that means a cloud that is not accessible via a smartphone? We must not forget that smartphone operating systems like Android were never designed to enable business-to-business transportation systems. In the various versions of Android there are an incredible number of loopholes that have still not been closed. That is why you must not link safetyrelated systems in a vehicle with any infotainment and navigation systems. Could you please give an example? First of all, the HD map is nothing more than a data cloud consisting of individual data points, which can be equally used as input for infotainment systems as for the development of driving strategies for highly automated driving. These applications must simply be run separately from the earliest possible stage. It must be impossible for a consumer navigation app or a service app to interact with the vehicle system in such a manner as to gain access to safetyand driving-related areas. We at Here already have comparatively many years of experience with such business-to-business IT systems, as we call them, because we provide similar products to the public sector, to governments and administrative departments. For us, one thing is clear: The solution can only be a standalone cloud for mobility applications such as highly automated driving. In other words, what we heard some time ago about seizing control of automotive electronic systems was, in a sense, a shot across the bows? You are right. I guess that was just the warning shot that we needed, because in the past five years we have seen tremendously rapid growth in the field of vehicle connectivity, and the system security functions really need to keep pace. But development is continuing apace. In the automotive sector we have the relatively secure GSM/ UMTS air interface, which is being further refined precisely because of constant attacks by hackers. At the same time, operating routines must be developed that are able to detect an abusive intrusion and react in a way that will prevent anybody from getting hurt and also minimize potential economic losses. I’m convinced that it is possible to place safety-relevant and personalized data behind a firewall that will be secure enough to create one of the safest IT systems anywhere - and we are talking about a system that, in 10 or 15 years, will be used in millions of vehicles across the world. So let’s just keep looking into the future for a moment. Can highly automated driving already be implemented in practice? Technically it is doable. The open questions relate more to regulatory issues, i.e. the political and normative standards, and to its economic viability. Highly automated vehicles are not built for their own sake. The end consumers really have to find such a system so attractive that they are prepared to invest more than they would for a conventional vehicle. Furthermore, we must not forget that all these vehicles have a certain lifespan. They will have to be constantly updated over their lifecycle in order to reflect current state of technology, safety and security solutions, and at some later point, they will be discarded. New service structures must be created for this, which on the one hand takes time and on the other requires a certain level of market penetration. That means that we will see a gradual development in market penetration, via several degrees of autonomy, until truly autonomous driving will be achieved far in the future. Actually, I do not like the term ‘autonomous driving’. In practice a vehicle is certainly not able to act completely independently, i.e. ‘autonomously’ in the literal sense. It is always dependent on the infrastructure and must obtain information about the road conditions and its surroundings. Nevertheless, there are more and more ideas for corresponding uses around, such as autonomous mini taxis picking up the elderly to take them shopping. How realistic is that? The advantages of such vehicles would be enormous, particularly in an aging society, because such services could reintegrate a significant share of the population into social life. In a completely measured and charted environment - such as in Cupertino in California - that is actually working well. However, in the normal road traffic in today’s cities, I consider the prospect of any general deployment of such applications to be science fiction for the time being. ■ BACKGROUND INfORMATION About Here Here is a Nokia subsidiary. It brings together the expertise in the fields of mapping and development of location-based services that Nokia has systematically built up since the mid-2000s. Some of the decisive steps in its development included the acquisition of gate5, a Berlin-based navigation start-up, in 2006; of Navteq, which was founded in 1985 and is the world’s largest developer of digital map data, in 2008; and of earthmine, a specialist in 3D cartography and street panoramas, in 2012. Here has existed as an own brand since the end of 2012 and supplies customers from the automotive industry, the enterprise sector and the retail business with maps, locationbased services and location intelligence. In August 2015, Nokia announced its intention to sell Here to a consortium consisting of Audi AG, BMW Group and Daimler AG. The sale is expected to be closed in the first quarter of 2016.