Where is the German Autonomous Car? – Part 1: Myths

If you believe the media, 2023 was a major setback for autonomous cars. Cruise, a prominent company, had to temporarily suspend the operation of its robotaxi fleets in the USA due to a crash. But there is no denying that autonomous cars are a reality. Waymo, which began developing autonomous cars as a commercial product in earnest in 2009 at Google as Project Chauffeur under the leadership of DARPA Grand Challenge winner Sebastian Thrun from Solingen in Germany, now offers a commercial driverless robotaxi service in several cities in the US. Cruise also plans to start operating again in a US city later this year, but this time more cautiously and with the involvement of local residents.

In addition to Waymo and Cruise, the next companies are already at the starting line. Amazon’s Zoox, WeRide, AutoX and Apollo/Baidu are just a few of the names, some of which are backed by financially strong companies. The latter companies originate from China, where a dozen companies in several cities and regions are also striving to bring robotaxis and self-driving technologies to market maturity and approval.

However, efforts are most advanced in California, where the scene’s attention is also focused. This blog, with a book of the same name in several translations, has been observing this scene for almost a decade. I myself have had access to the first robotaxi fleets as a passenger for a year and a half and have now ridden in driverless cars more than 170 times. I have also taken more than 350 people on these joyrides so far. The majority of my passengers have come from German-speaking countries and the next ones have already been announced.

I have also been able to drive vehicles from a number of other companies and gain an initial impression of their development status. And just by the way, I’ve also had the Full Self-Driving Beta (FSD Beta) on my Tesla for about a year and can see its progress, but also occasional setbacks.

I am involved in the development of autonomous cars not only because, as a former SAP developer for many years, I find the technology exciting, but also because visitors from German-speaking countries in Silicon Valley have repeatedly made similar false claims about electric cars and autonomous vehicles, which are still widespread. In the meantime, electric cars have sobered up and Tesla is no longer just seen as a crazy company that will soon go bankrupt, but has shaken up the automotive industry worldwide and forced it to make changes. Here, traditional companies find themselves in a state of cat’s meow, trying to catch up with what needs to be caught up with, but the results are mixed at best.

Electric cars have now been replaced by autonomous cars as the content of these false truths. Similar (false) arguments against autonomous cars and opinions about their manufacturers are being put forward from Germany, many of which are based on false assumptions. At the same time, people are telling themselves that Germany is much further ahead with autonomous cars, but that this technology is being developed by German companies in a way that is less market-hyped, safer and, above all, more perfect than by the ruthless and law-breaking Chinese and American companies.

But is that true? Let’s take a look at a few popular myths, but let’s start with a question:

Have you already encountered autonomous cars on German roads? Maybe even a robotaxi fleet?

I always ask this question at presentations and in discussions with participants in delegations and expert panels. In 99 percent of cases, the answer is no. Even people who work in the automotive industry have never or very occasionally seen such vehicles on German roads, at best once at a trade fair as an exhibit. The few who have seen one have encountered the vehicle on the highway. But no one in Germany or other European countries has ever seen entire fleets.

Visitors to San Francisco, on the other hand, cannot walk through the city without repeatedly encountering autonomous test vehicles from various manufacturers – some of them without a driver behind the wheel. That’s often the first thing they mention to me. Quite clearly, because I often meet them to talk about exactly that and to call and drive such a vehicle with them myself.

But what does that tell us?

I often hear and read a lot of myths from the German side, I am directly presented with false claims, which are rarely questioned and which often raise no doubts at all about one’s own perception. Why is that the case?

There are a number of reasons for this. Let’s go through the myths, the questions you should ask yourself and the reasons (I’ll cover the last two in part 2 of the blog).

Myth 1: Only with Connectivity

Claim: Before autonomous cars can function, roads must first be converted into smart roads. This would absolutely require 5G and sensors in the roads.

Fact: California shows that it can be done without them, as there are no standards for communication between vehicles and road infrastructure (V2X) or between vehicles themselves (V2V), nor can we rely on nationwide 5G. Especially in Germany, where in many regions you are lucky if you see an E on your cell phone.

Companies would of course use broadband if it were available. But since it’s not available nationwide and reliably, autonomous cars are being developed without the mandatory function of being able to communicate with others. And that makes sense, because every channel opens a gateway for cyberattacks. But a hacked autonomous car that can be misused as a two-ton weapon, and perhaps even in fleets, is the last thing we want.

Also, on closer inspection, there are hardly any scenarios where communication is absolutely necessary. Almost every scenario put forward by connectivity supporters can be solved differently. Another argument against connectivity is the cost of the infrastructure. Which road operator wants to voluntarily spend money on this? Especially when autonomous cars open up the possibility that road signs or traffic lights, which are used for communication between the road infrastructure and human drivers, can be almost completely removed? Three million traffic lights, 20 million road signs and 4 million signposts in Germany cost a huge amount of money. Digital road maps and negotiation on sight with priority rules at junctions – as is already the case in shipping – make this infrastructure largely superfluous.

Myth 2: Test Tracks and Simulations Make Perfect Cars

Claim: German manufacturers test the cars so thoroughly that they will hit the roads almost perfectly. They do this responsibly and not in such a risky way as American or Chinese manufacturers.

Fact: Former Waymo CEO John Krafcik revealed in an interview years ago that 80 percent of the development progress in autonomous driving is made possible by simulations. Around one thousand times as many kilometers are driven in the simulator than in reality. In the very realistic simulators, real driving scenarios experienced by the test vehicles can be played in, simulated and thousands of variants tried out.

But we know that reality is more complex than it can be recreated on test tracks or in simulators. That’s why there comes a moment of truth when you have to get out of the simulator and off the test tracks and drive in real traffic. No matter how much imagination you can muster, no test scenarios encompass the complexity and craziness of reality. I put together a collection of videos of various real-life scenarios a few years ago. Cruise even published such a summary with a matching song.

As is well known with any plan, no matter how good, as soon as it comes into contact with reality, it becomes meaningless. Heavyweight boxer Mike Tyson responded to a reporter’s question about his plan for his upcoming boxing match against Evander Holyfield:

Everyone has a plan until they get punched in the mouth

We know from painful experience that new technologies like an airplane can’t just be developed in a wind tunnel and a simulator. The real test comes when the airplane makes its first test flight. After more than 100 years of human flight and countless accidents with fatalities and injuries, flying is one of the safest means of transportation. But until we got there, there was no detour.

Now many are arguing with the fervor of conviction that we can’t afford that with autonomous cars. and they’re right, but only partially. Nobody wants to see as many deaths over the decades from autonomous cars as from airplane accidents. At the same time, we know that one million people die and 10 to 12 million are injured in road accidents worldwide every year. Time is of the essence to replace human drivers with safer technology.

But if we apply stricter standards to autonomous cars than we do to human drivers, then we are accepting that a large number of people will be unnecessarily injured on the roads for longer. If we were to follow the most ardent advocates of absolute safety requirements, we would still not have commercial air traffic today, but would still only have flying in test mode.

The fact that you don’t see German manufacturers’ vehicles on public roads is an important indication that they are far behind in terms of technology. And that leads us straight to the next myth:

Myth 3: German Companies are at the Forefront!

Claim: German companies are – according to German experts – if not leaders in autonomous driving, then they are at the forefront.

Fact: Based on the points mentioned above, you can no longer claim or believe this with a clear conscience. Above all, it sounds as if you are trying to encourage yourself. But if someone does this despite the facts mentioned, then there is a hidden motivation behind it. It is a case of lulling and deflecting, in which consultants and experts in the automotive industry do not want to alienate their (still) solvent customers. A tactic that serves self-interest. But we have to stop the whitewashing and ventriloquism and talk straight. The truth is unpleasant, and we have done nothing for too long because we were lulled into a sense of security.

If German manufacturers were at the forefront, they would have to be seen as Waymos & Co. In fleet operation as robotaxis with external passengers. Or be available as an option in cars sold today. But where are they?

The development gap between German companies and the leading companies such as Waymo is calculated: they started practical drives very early on, which picked up speed from 2015 to 2017. This puts the development gap at 5-7 years. An eternity in the software world.

Myth 4: Responsible Germans and Lawless Americans

Claim: Americans are reckless, Germans are very responsible and do it right the first time.

Fact: If you know the legal system in the USA, then you know that the penalties there can be much higher and more painful than in Germany.

As soon as the conversation turns to the USA and Americans, or even China, and their autonomous cars, the argument is inevitably put forward that German car companies are much more responsible and safety-conscious than the Americans and Chinese. After all, an accident involving an autonomous car would be a PR disaster for companies like Mercedes-Benz, whose brand stands for safety. So they couldn’t afford it.

The Americans and Chinese, on the other hand, are lawless and nefarious and risk the life and limb of others for their progress. You can’t simply disregard laws.

All I have to say is: Diesel Scandal.

And German laws are almost toothless at this point. While the companies and those responsible have not even really received a slap on the wrist, fines in the billions have been imposed in the USA. California alone fined VW two billion euros to build a charging station network.

Myth 5: Americans are Better at Marketing

Claim: Americans have a bigger mouth, they are better at marketing. On the other hand, German companies are cautious and create the big things in the background!

Fact: The best advertising for your own autonomous cars is what? When you see them driving around in public. Do you see the American and Chinese manufacturers driving in many cities and regions in the USA and China? Yes! After all, this blog is a testament to that, where I’ve encountered hundreds of these vehicles in the US, filmed them and published the videos here.

Do you see German autonomous cars in German cities? Not really (although I have published a small single-digit number of pictures and videos over the years, often sent in by readers).

The marketing comes automatically when you let the vehicles drive in the real world. So German manufacturers are no worse at marketing, they just don’t have anything to show off that could attract marketing attention.

Myth 6: Tesla Autopilot

Claim: Tesla Autopilot and the many accidents with fatalities clearly show that autonomous driving does not work.

Fact: Tesla Autopilot is not a technology for autonomous driving (i.e. level 4 or 5), it is a level 2 driver assistance system in which a driver still controls the vehicle at all times.

The fact that the autopilot has prevented accidents in many cases is often left unmentioned or not even known. However, the latter are less spectacular than those where an accident has occurred and many media report on it. It is also often overlooked that many of these accidents attributed to Tesla Autopilot turn out in retrospect to be the fault of the drivers and that they were lying. One of the best-known cases dates back to November 2022 and involved a Tesla driving through Chinese streets at speeds of up to 164 km/h, resulting in two deaths. As the analysis of the driving data revealed, the driver had mistaken the brake pedal for the accelerator pedal, pressed it down 100% and moved the steering wheel. The Autopilot was therefore not to blame.

The Tesla Autopilot should also not be confused with the Tesla Full Self-Driving Beta. These are two different functions. As already mentioned, the former is a driver assistance system, while the latter aims to reach Level 4 functionality, but is currently only legally declared as Level 2 by Tesla. Incidentally, the FSD beta is currently only available for testing on almost 400,000 Teslas in North America, but not in Europe.

But as soon as autonomous cars are mentioned, German-speaking participants in the discussion immediately refer to Tesla Autopilot, usually in a negative way. I have already discussed this “German fixation” on the Tesla Autopilot and identified it as a mentally blocking problem.

In most cases, Tesla Autopilot is the only technology that Germans think of as self-driving technology and autonomous driving. However, Tesla is neither a leader nor is its approach (cameras only) comparable to that of the leading company Waymo (use of cameras, lidars, radar, etc.). I have already explained the difference in other blog posts and in this blog post.

So anyone who mentions Tesla Autopilot in a discussion about autonomous driving reveals themselves to be someone with little knowledge of this technology.

Myth 7: Crashes and Incidents

Claim: Autonomous cars often have accidents and constantly break down.

Fact: Several studies (for example here and here) indicate that autonomous cars and robotaxis already have significantly fewer accidents and that far fewer people have been injured in incidents than by human-driven vehicles. However, the few incidents that do occur attract the most public attention. Just as the risk of fire from batteries is played up in the case of electric cars (in Germany alone, around 60 petrol and diesel cars burn down every day – no wonder: they are brought to a controlled explosion with a flammable and highly explosive liquid in a “combustion engine”), incidents involving autonomous cars are being played up.

The California Department of Motor Vehicles (DMV) requires holders of an autonomous driving license to submit a report of every collision that has occurred. The DMV publishes the list of reports on its website. If you analyze these reports, it quickly becomes clear that more than 95 percent of collisions were caused by the other party, always a human driver.

With the increasing number of autonomous cars in the form of robotaxi fleets or software, some of which are even in a beta version on customer vehicles (see also the Tesla FSD beta), there is inevitably an increase in incidents. The reports range from accidents and awkward situations to vehicles stalling. Due to the number of vehicles from one manufacturer, we are also observing fleet effects for the first time. Incidents that do not occur with individual vehicles, but only in a fleet, such as a dozen vehicles stalling at a junction. This reminds me of my time as a software developer, where you test your code with a test data set of a few thousand lines of data, but as soon as you run millions of data sets through the code, errors occur because it contains data that you didn’t come across in this way or in this combination during programming.

Compared to the many incidents and violations of traffic rules by human drivers that are commonplace and neither noticed nor reported, those of autonomous cars are negligibly rare. However, it is precisely because they are new and so rare that they receive this media attention in the first place.

And they also receive this attention in Germany, where reports from the USA are often reported in a smug tone, making people feel good because it shows that “the Americans can’t do it either” and “it won’t work”.

Conversely, are there any incidents from Germany that have come to our attention? Not really (with perhaps one exception where BMW was involved in a fatal crash with a test vehicle). But that’s not surprising, because where there are no autonomous vehicles, there are no incidents caused by them, and certainly no fleet effects.

Myth 8: Won’t Come for Another 10 or 20 Years, or Never!

Claim: Autonomous driving will only come in a) 10 years, b) 20 years, c) never.

Fact: People who firmly believe this are right. This technology will only come from them and their companies in 10 years, 20 years or never. But it is already here today from American and Chinese companies and is being improved.

The best sign of this is the news that Bosch plans to cut over a thousand jobs worldwide at the beginning of 2024, particularly in those departments that develop software for autonomous driving.

Myth 9: Unfortunately, the Laws Prevent it

Claim: We would love to, but unfortunately the legal situation in Germany prevents or hinders development in the country.

Fact: Germany has had a law on autonomous driving since summer 2021, and even before that, car manufacturers were able to test prototypes in public without a major approval process. But let’s get into the details.

One of my favorite myths, because on the one hand, the (lack of) laws prevent the development of autonomous cars in Germany, while on the other hand – as we have already heard – we are at the forefront.

Until 2021, the lack of a legal framework for autonomous driving was put forward as an argument for the backlog in the development of self-driving technology in Germany and that unfortunately nothing could be done as an automotive industry. This feigned impotence of the most powerful industry in the country is an act, because in other matters – such as the delaying of EU-wide directives on emissions standards – it was easy to get the then Chancellor herself to intervene in Brussels.

It is quite clear what was behind this: delaying excessively strict emissions standards was important, a legal framework for autonomous driving was not.

But this excuse no longer applies, because in summer 2021, Germany became the first country in the world to introduce a law that enables autonomous driving. Have fleets of autonomous cars appeared on German roads since then? Of course not, because on the one hand you would have to have more mature technology, and then the local and federal authorities would have to know who is responsible for approval and what needs to be assessed. The kind of experience that California, for example, has amassed over the last 15 years is something that you first have to gain yourself in order to know what requirements applicants have to meet in order to obtain a license. Questions such as “How do I actually test the safety of a self-driving system?” or “Do emergency personnel have to be trained on autonomous cars, and if so, what does the training involve?” have to be created and survive the practical test.

Myth 10: Nobody Needs Autonomous Driving!

Claim: Autonomous driving is a solution in search of a problem. Nobody needs it.

Fact: There are a number of good uses for autonomous driving and reasons why it is being developed. First of all, human drivers cause many accidents every year, resulting in many deaths and injuries. Developing a technology that helps us stop drunk, distracted, tired, or angry people from causing accidents should be close to all of our hearts. Furthermore, autonomous driving technology allows the creation of completely new types of vehicles, such as the self-driving supermarket from start-ups like Robomart.

The shortage of drivers is also a major problem in many sectors today. Truck drivers are desperately sought after, cab drivers are getting older and older on average, as are driving instructors. At the same time, the number of young people getting a driver’s license is decreasing and, if they do, they do so much later and rather listlessly. Transport companies are not spared from this either; the lack of personnel is thinning out the timetables of entire lines.

But why do we hear this argument so often? It has to do with the fact that automotive companies have a self-selecting process. People who apply there are interested in cars and like to drive vehicles themselves. This means that these companies mainly employ people who enjoy driving. This makes it difficult for them to imagine that there are people who don’t like driving, refuse to drive or are sometimes temporarily unable to drive a car themselves. And yet they would like to have access to individual mobility.

For me, a visit to the French manufacturer of autonomous shuttles, Navya, in Lyon was eye-opening. My contact there told me that only 40 of the 220 employees had a driver’s license. These employees develop autonomous mobility for themselves.

Myth 11: Autonomous Cars Are Ugly!

Claim: The autonomous cars from Google & Co. are ugly, they have no idea about design!

Fact: Cars that have entire sensor superstructures on the roof and sides certainly don’t look attractive. Even the Google Egg, or Dragonfly (as it was known internally) or Koala, was more round than beautiful. And I wrote a detailed blog about this back in 2016, because an autonomous car, being the robot that it is, automatically has to have other design elements. In such a way that the sensors don’t have any blind spots and users don’t feel threatened, but that it appears friendly to them, not only in terms of its appearance, but also in terms of its behavior. If we also assume that it is not just a linear continuation of traditional mobility behavior (individuals buy a car and drive it), but one where sharing systems dominate, then the appearance and color of the car, like a cab, is quite unimportant.

To summarize it in a nutshell, in car companies 30 to 50-year-old men build cars with which they can impress women. Autonomous cars, however, will predominantly replace private cars with hard systems such as robotaxis in fleets, and so a sleek design will become less important, but one that people perceive these robots as friendly.

Myth 12: People Prefer to Drive Themselves

Claim: People want to drive themselves and not give up control. They also like driving cars.

Fact: It is true that there are some people who like to drive themselves, who enjoy being in control of the machine. Some are also afraid of entrusting themselves to a machine (or another driver) and imagine the worst risks.

It is also a fact that these arguments always come from people who have never driven an autonomous car themselves. The consistent feedback from the now 350 or more passengers I took on my rides was always that they were amazed at how normal it felt and that they couldn’t wait to have autonomous cars in their homes. After just one drive, they were ready to hand over the steering of their own vehicle to the machine on a regular basis.

And when someone tells me that they like driving their own car, I simply ask them to tell me again the next time they are stuck in a traffic jam how much they love driving their own car.

And yes, of course it’s nice to take a nostalgic ride in a steam locomotive, or go horseback riding, or send a postcard, but for daily errands, commuting and work, we prefer modern means of transportation or communication. At the weekend, we can always take part in a classic car rally on the Grossglockner or a coastal road on Sylt and drive a vehicle ourselves – on a closed-off stretch of road, of course, because anything else would be too dangerous and reckless.

In the second part of this series, I will look at further questions and reasons why no autonomous cars are coming from Germany.

But first, a reference to my latest book, Creative Intelligence, which deals intensively with generative artificial intelligence.

This article was also published in German.

Read More