Car manufacturers are continuously making cars ‘smarter’. Apart from adaptive cruise control, cars are also capable of parking independently. This article takes this development one step further, what if the car could drive without its driver? Although autonomous cars may seem science-fiction, the realization is well within reach. This article will list the main advantages, elaborate on the levels of automation as well as highlighting the technological aspects of autonomous cars. Then, regional differences will be analysed in order to identify the regions with the highest growth capacity. Obviously, the main challenges have to be introduced as well, before the article will be concluded with an estimated timeline of adoption. In theory, autonomous cars have the force to disrupt the entire automotive industry but will autonomous cars find the momentum necessary to penetrate the industry fully?
Why go autonomous?
Understanding the advantages of autonomous driving is crucial for a clear comprehension of its potential disruptive impact. In order to make this abstract topic easier to comprehend, let me illustrate this.
Imagine, it has been a long and exhausting day of work and you want to call it a night. However, high priority to-dos remain to be done before you can leave the office. Instead of lingering back to your desk, you summon your car via your phone. Just while leaving the elevator, you receive a confirmation that your car is ready to go. The car takes off while you got plenty of time to finish the last pressing matters. While preparing your meeting for tomorrow, you get a video-call from a colleague in Tokyo who asks you to run over a few documents with him. When you get home, all to-dos are coped with, tomorrow’s meeting prepared and the car’s pick-up time is programmed, so you can devote all your attention to your family.
In order to provide this seemingly far-fetched story with a somewhat more theoretical basis, I will elaborate on the main advantages.
Autonomous cars, if implemented fully, will prove a real lifesaver. In the US alone, 11 million people have been killed in car accidents, mainly due to disobeying traffic rules, fatigue, driving under influence (alcohol and drugs) and incompetent drivers. All these accidents can be prevented by technology; cars will not get tired, do not drive while under influence and are more competent than any human driver. This reasoning could be taken one step further, because if cars don’t crash, why do we need all the features that make cars safe, like heavy metal chassis?
Next, autonomous cars can make driving so much more economic in terms of fuel usage. It is estimated that driving on cruise control, which autonomous cars can do continuously, can save approximately 15-30% on fuel. When we reach the further stages of development, technologies like Vehicle-to-Vehicle (V2V) and Vehicle-to-Infrastructure (V2X) communication, will allow cars to anticipate even more accurately. Extrapolating these saving rates to the annual national oil consumption in the USA (more than 140bln gallons of oil), the implications for the economy and environment are disruptive, to say the least. Not to mention, the impact on the global economy.
Based on the same technologies (V2V and V2X), traffic jams will be prevented. This will not only convey higher time efficiency but also fuel savings. In the US alone, 3 billion gallons of oil is wasted annually through traffic jams. The main factors causing traffic jams are: accidents, rubbernecking (drivers watching accidents) and too high traffic volumes. As established earlier, accidents will be ruled out totally, as development reaches the later stages. Consequently, rubberneckers will have nothing to stare at anymore, eliminating two of the three main reasons listed above. Traffic jams due to under-capacity of some very busy roads will be solved as well by relying on V2V and V2X systems. These systems will redistribute all traffic evenly over all the different roads, thereby creating a much more even flow of vehicles.
As indicated in the example of above, the freed up time created by not spending time driving is a major advantage. Whether one chooses to work during the entire commute or to relax, the utility to be gained here is substantial. Estimating the magnitude of these gains remains difficult though, since so many factors are still uncertain. This also affects the division on the value of a car (see figure 1). Historically, hardware took first place but with software and infotainment in place, this takes a substantially larger scoop of a car’s value. However, the bottom line is quite clear, consumer productivity and utility will increase due to autonomous cars.
Figure 1: Value of the car – Today vs. Tomorrow
Source: Morgan Stanley Research
The impulse that autonomous cars can have on the economy is a very broad topic. I will highlight some of the advantages to the economy. As mentioned previously, consumer productivity will increase, decreasing the slack in productivity due to having to drive. The hardware and software sector will thrive because of the need for new software and hardware systems, driving autonomous cars. Moreover, people have more time to consume, whether this is via e-commerce or traditional shopping methods. The advertisement industry, too, will see increases in demand for advertisements, such as billboards or in-car entertainment. Since all autonomous cars are connected to a network, location based advertisements have serious potential as well. Next, the long-haul freight industry may benefit greatly, since obliged rest times drag down efficiency. This can be effectively solved by autonomous cars, as they are not restricted by such limitations.
Levels of Automation
A potential disruptive trend like autonomous cars is not one that presents itself instantaneously; this is something that develops over time from smaller break-troughs. In order to get an overview of this development, it is best to scrutinize these trends individually.
IHS Automotive indicates five different stages of development (see figure 2). The current developments mainly involve Advanced Driver Assist Systems (ADAS), like Adaptive Cruise Control (ACC) and autonomous breaking systems. These systems occur in phase 1, which we are currently in. The high-end spectrum of the automotive industry already encompass a lot of these features, while these developments still have to reach the lower segments of this branch. Costs will decrease when these techniques are implemented more frequently, which is a key element for further expansion and implementation of the upcoming technologies. The ADAS industry is relatively small currently ($1-2 bln), but Citi estimates that this will reach $40 bln in the later stages of development. This amount, if true, carries serious weight for the companies active in this industry, Citi identifies the most influential companies to be the following: Continental, Denso, Autoliv, Delphi, Bosch.
Figure 2: The dynamics of autonomous driving
When these assist-systems are implemented widely and fine-tuned, they eventually reach a phase in which they are so accurate in anticipating unforeseen events, that limited self-driving is the only logical next step. Following the same reasoning, semi-autonomous cars are introduced, which form the last barrier before going fully autonomous. As can be seen in figure 2, reaching these later phases of development will take some time. Specifics on the precise timeline of adoption will be given later on in this report (Timeline of Adoption).
Scepticism exists too, as ADAS systems are the frontrunner for full implementation of autonomous cars. This concern mainly revolves around the question of: why do consumers pay high amounts for monitoring systems, if they still have to monitor themselves? The early stages of development are never perfect but are necessary to take technology to the next level and since most of the ADAS are marketed based on safety, I believe this is not insurmountable. However, it is critical to have a clear picture of all the challenges threatening this trend, which will be discussed at a later point in this article (Challenges).
Technologies; who wins, what are challenges?
As stated earlier, autonomous cars seem fiction, but the near-term potential is regularly underestimated. Nearly all the hardware, such as sensors, radar systems, and networks, exists already. Yes, the costs of this hardware are still relatively high (about $10k for full features) but once implemented more frequently, average costs are estimated to decrease due to economies of scale ($5k). So, if hardware is not the limiting factor on the implementation of autonomous cars, what is? Most of the pertinent issues revolve around software (see figure 3), especially the algorithms needed for analysing all of the data that flows into the car’s systems. Furthermore, in order to fully realize the opportunities of next-generation ADAS technology, future cars will require much more computing muscle. The computing power of current cars is still relatively limited. Chip manufacturer Intel, estimates the inflow of data into semi-autonomous vehicles to be around 1GB per second, which all needs to be processed extremely fast. Thus future cars will need to be equipped with far more computing power.
Figure 3: The necessities to go autonomous
Source: IHS © 2014 IHS
Then, there is a choice to be made on programming the cars to recognize patterns and to actively learn from its surroundings. The pattern recognition challenge is mainly based on communication with infrastructure. Solving this problem is much cheaper than the second issue of creating new adaptive learning algorithms, this is because such communication technology requires less hardware and software to be developed. However, the initial costs are high, due to large investments necessary to modify all current infrastructure. Moreover, infrastructure in Developed Markets (DMs) is quite easy to modify, but this is not so easy in Emerging Markets (EMs), thus increasing the barrier for the overall trend to penetrate fully (Regional Differences will be discussed in the next section).
That is why most pioneers are developing independent systems, which analyse information from on-board resources to get a 360° view of the surroundings. Although the overall difficulty of developing such a form of autonomy may be higher, implementability in EMs and not having to re-modify infrastructure constantly are big plusses.
A big concern is whether the early adopters can generate enough volume for the Original Equipment Manufacturers (OEMs) to invest enough resources into the R&D of autonomous cars. Not only the Return on Investing (ROI) in these technology matters, but also the incremental price change this involves for the price of the car. Fortunately, outsiders like Google, IBM, Cisco and smaller start-ups do invest heavily in the development of autonomous cars, which pressures OEMs to keep up.
Still, the automotive industry remains a conventional, traditional and high-quality sector, based on customs which have been present for several decades. However, if these OEMs want to preserve this status, they will have to exchange some convictions for out-of-the-box thinking. Failing to do so will have severe consequences for these OEMs in the medium term, while greatly slowing down the penetration of autonomous cars.
In order to get a solid overview of this trend, the geographical factors have to be analysed.
On the widest geographical zoom, DMs appear better suited to embrace autonomous cars than EMs.
One of the reasons for this is that DMs have more advanced infrastructure and networks than EMs (see figure 4).This is important since even the most advanced and independent autonomous cars need information from infrastructure and need to be connected to a fast and stable network, such as car-to-car communication systems like Dedicated Short Range Communication (DSRC).
Figure 4: Infrastructure quality vs. Traffic deaths
Source: Euromonitor Data, Morgan Stanley Research Source: World Health Organization - Note: 2012 data Note: Light vehicles excluding two wheeler
Moreover, DMs most probably have a higher willingness and ability to pay for these additional luxury features. Obviously, automotive industries in EMs are objectively less developed. However, EMs also prove a very interesting market for autonomous cars. Firstly, EMs are way less regulated by the government, which mitigates a major drag for the implementation of autonomous cars. Moreover, road safety in EMs is distressing to say the least, autonomous cars will be a perfect solution for solving this. However, autonomous cars need to be able to navigate through highly chaotic situations, which pressures the algorithms even more. Therefore, the main consensus is that although EM adoption is important for full penetration of autonomous vehicles (see Challenges), DMs will likely be the first to implement autonomous cars.
In order to make this statement more concrete, I will analyse the two most forthcoming regions: Europe and North America. Europe is expected to be a frontrunner, based on its high quality automotive sector. Strikingly, this is not the case. As a matter of fact, North America currently seems most willing to pioneer. This may be because North America’s economic environment is very welcoming to technological initiatives (Silicon Valley) or because they want to achieve a higher market share of the Europe-biased automotive industry. Two states (California and Nevada) in North America have already given out the first permits for autonomous cars to actually be tested on public roads. Moreover, these two states have already started addressing one of the most pressing concerns for autonomous cars: liability issues (see Challenges).
It is of upmost importance that Europe does not lose its first-mover position in the automotive market by focusing too much on current cash cows. This short-term strategy could rapidly turn into a nightmare, when they realize that they have been overtaken because they cannot adhere to 2025+ standards.
As stated earlier, autonomous cars can only reach a state of full penetration when the whole world accommodates this. Although it may seem tempting to exclude EMs from this development for cost reasons, this probably will backfire on the entire industry. Fully functioning autonomous cars can be much lighter than current cars and, more radically, may look nothing like the car we know today (see figure 5). This implies, that if autonomous cars are not or cannot be implemented in EMs, OEMs need to have two full production lines, increasing the costs of production drastically for both type of cars. Furthermore, continuing production of ‘traditional’ cars will further slowdown the replacement process of substituting regular vehicles for autonomous cars. In short, it is very important that regional differences are continuously analysed, in order to identify local pitfalls and create innovative solutions that might improve technology and adoption for both DMs and EMs.
Figure 5: BMW Concept Car
The seemingly unrealizable status of autonomous cars can only be proven wrong if all challenges present are eradicated effectively. That is why this section provides an extensive overview on all challenges.
Frequently, the first question on autonomous cars concerns liability issues; who or what is responsible when the equipment fails and causes an accident? It is extremely important that conclusive legislation is developed in order to solve these delicate questions (see figure 6). A related issue concerns ownership questions on data created by autonomous cars, since this data is not only valuable, but also private and potentially revealing. Although liability issues are important, similar challenges have never stopped human kind to revolutionize (cars, plains) and therefore I believe these challenges are not insurmountable.
Figure 6: Legal issues to be addressed
Source: IHS © 2014 IHS
Following down the issue on data generation, another major concern as with any technological trend these days, is cyber security. Autonomous cars will almost certainly be the subject of cyber-attacks.
This likelihood only increases when cars are connected to various types of networks, necessary to operate systems, like DSRC. Moreover, with the market for infotainment increasing due to not having to drive, hackers have a wider array of hardware to hack. Cyber security plays an even more important role than for other trends, such as Big Data, because cyber-attacks on autonomous cars might directly influence the loss of human life. Key part of a solution is that unauthorized events must be detectable. This may sound redundant but current hardware is often not able to detect breaches. Furthermore, the cyber-security market needs to address these safety questions adequately. As it is almost certain that these attacks will happen, let us at least make sure that we can identify and handle them.
Another challenge for autonomous cars is based on consumer acceptance, which closely conjoins with the cost of this additional feature. Given the sci-fi status and the previous concerns on liability and security, consumer acceptance may actually be strenuous to achieve. As with every trend, frontrunners are needed to literally advance and develop the feature. However, for full penetration of autonomous cars, these early-adopters do not even closely resemble the necessary user volume. As noted earlier, cost issues play an important role when the feature has to be paid for but still requires the driver’s monitoring. Currently the cost to add autonomous features is around $10k, which will decrease steadily as this trend matures. It will take time for people to put one’s safety in the virtual hands of a robot. However, as robots will play increasingly more roles in our daily lives, this is something human kind inevitably has to get used to.
Replacing the current car parc is also crucial for the implementation of autonomous cars. However, when analysing the size and the replacement speed of car parcs around the world, this may take quite a while. Currently, the American car parc counts more than 253 million cars and 13-14 million cars are wrecked (Morgan Stanley) annually. This implies that replacing the total American car parc alone will take approximately 20 years. So full autonomous penetration requires that all cars replaced for the next twenty years are purely autonomous. This fact indicates that the ‘utopian state’, or full penetration, is likely to happen on a rather long time scale, which is something for investors to bear in mind.
Possible Timeline of Adoption
In order to place autonomous cars in its respective time frame, this part will study the timeline of adoption. As discussed previously, autonomous cars are closer than people think, but how close is this trend really? Opinions differ on this point, OEMs estimate it to take longer than a decade while outsiders like Google (and Tesla) aim for 2020. Morgan Stanley (MS) too follows a rather bullish approach by estimating that semi-autonomous cars will take 12-18 months and fully autonomous cars will exist before decade’s end (see figure 7).
Figure 7: Possible Timeline of Adaptation
Source: Company Data, Morgan Stanley Research
At present, OEMs are modifying current cars with ADAS. In the future, they will have to adapt by designing specific autonomous cars to exploit all gains from autonomous driving. Outsiders of the automotive market follow a very different approach by directly designing and producing the car of the future, i.e. dedicated autonomous cars. New entrants to the automotive market have to follow this strategy since they cannot capitalize on this semi-autonomous phase. This yields two distinct advantages. First, giving people all the advantages of autonomous driving is the most promising approach for full penetration. Secondly, outsiders want to monetize fully on the new potential revenue flows from infotainment and other on-board equipment by totally re-inventing the vehicles. This is something OEMs should keep a close eye on, because if they keep focusing on current designs they run the chance of totally missing out on these additional revenue flows, with potentially disastrous results.
Still, OEM market leaders in the automotive industry did not achieve these sales volumes by missing out on trends. Currently the OEMs in the lead are mostly German manufacturers like Audi, Mercedes-Benz and BMW, closely followed by Volvo and General Motors (GM).
Apart from OEMs, Google has already working prototypes of autonomous cars and is aiming for a piece of the pie. As noted, OEMs and outsiders are striving to gain the lead in developing fully functioning autonomous cars for everyday use. Probably, the first fully autonomous cars will be sold somewhere from 2020-2025 but the real impact of autonomous cars will be around 2035-2050.
This article has shed light on the issue of autonomous cars. The main question resolved is whether autonomous cars will be a feature to existing cars or the future of the automotive industry. The advantages of autonomous cars mainly consist of safety, environmental and economic reasons. Autonomous cars will not appear instantaneously but will more likely develop gradually from current safety features towards (semi-) autonomous systems. Technological difficulties have been discussed as well, establishing that the main concerns are present in the software, rather than the hardware part of these systems. Developing these systems is not only influenced by technological issues, but also by implications of regional differences. Although DMs are deemed most likely as the biggest market for autonomous cars due to lower entry-barriers, EMs should not be overlooked considering the higher relative number of traffic deaths and the necessity to include EMs in order to achieve full penetration. Lastly, different views on the timeline of adoption have been weighed. While some bulls forecast the first autonomous models on the road before 2020, the main consensus is that this will be around 2020-2025.
Many factors have to be gauged in order to assess the magnitude of this trend. However, the consensus is that this trend is inevitable and that the only question remains on timing. The major impact of this trend is estimated to take place in later stages of development, i.e. 2035-2050. So, given the continuation of the current technological advancement, autonomous cars will be the future.
Enjoyed this article?
With your help, we can keep this website ad-free. Support Foresight Investor with a donation!
Karsten is currently studying finance and is a member of Risk. He organised the Risk Finance Symposium and invests on his own account. He specializes in equities, technology and emerging markets.