In the course of the Intelligent World Drive, the S-Class had toured cities including Cape Town, Melbourne, and Shanghai, meeting kangaroos, koalas, and “free-range” pedestrians, successfully crossing seven-lane intersections, and making hook turns. What could possibly shock it in L.A.?
After all, Los Angeles, Hollywood, Beverly Hills, Santa Monica, and Malibu are the world of the rich and famous, with chilled-out lifestyles, skating in Venice Beach, and strolling along the Walk of Fame. This isn’t a world of chaotic traffic… or is it?
Of course it is. Los Angeles is regularly stuck in traffic. A photo that went viral all over the world on Thanksgiving 2017 was intended to relieve Europeans’ self-pity over their own congested highways. It showed traffic at a total standstill on a twelve-lane highway in L.A. There are several deeply rooted causes for L.A.’s traffic problem: first, the urban settlement structure of this metropolis; second, the inadequate reach of the local public transport system; and third, the love affair of Los Angelenos, and Americans in general, with their cars.
As a result, Los Angeles is exactly the right place for our autonomous test drive in a specially equipped S-Class. Why? The first answer to this question comes from Frederick Kim, a futurologist at Mercedes-Benz Research and Development North America (MBRDNA). Los Angeles, he explains, is the epitome of the mega-suburb. It clearly exemplifies a trend that is equally prevalent in the vast majority of US cities. In short, forget about urbanization and gentrification, forget about mobile phones instead of cars, and forget about sharing rather than owning – at least that’s the current state for the USA. The reality is different here, say our futurologists.
In the USA, most young people
- are either moving to the suburbs of mega-cities or already living there (in L.A. the suburbs are especially sprawling due to the city’s geographic location, and they are relatively sparsely populated, with about 1,030 inhabitants /km2 compared to 3,008 in Stuttgart, for example),
- start families by having between two and four children (significantly more than in Europe) and own dogs (as a result, they need big cars),
- possess one or more cars, which is a requirement for daily life, as suburban distances are too great to walk, public transit is too inefficient, and taxis are too expensive. Many commuters to Los Angeles drive between 50 and 70 kilometers to work daily – one way.
California – a state full of German workmanship
This situation is especially striking in Los Angeles, whose public transport system is underdeveloped and plays virtually no role in the lives of many Angelenos. L.A. has developed around the car and has become the epicenter of car culture in the USA. The locals all drive – and they like driving. And there’s one more thing you can forget: In the ultimate Car City, people are not post-materialistic. Here the existential question is “to have or not to have.” Those who have money enjoy the good life, and those who don’t have it…well, of course the Uber drivers in particular have a lot to say about that.
Cars, especially premium cars, are still symbols of the good life. All the traditional attributes of cars – freedom, self-expression, and status – remain strong in the USA, even among millennials. This is good news for us, because in no other region of the country do you see more Mercedes-Benz cars than in L.A., the car capital of the USA. Los Angelenos explicitly appreciate “German workmanship” – especially the cars with the star.
Self-driving cars combat congestion
However, the result is that drivers don’t have anything like a free ride on the freeways in L.A. On the contrary, throughout the rush hour traffic stands still, even on six-lane highways. And traffic does not decrease significantly throughout the day, because when Americans are not driving to work or back they (especially the “soccer moms”) are chauffeuring their two, three, or four kids to school or to various recreational activities on a daily basis. It’s clear that driverless driving – think of robotaxis – could reduce traffic, and autonomously driving private vehicles would at least allow drivers to make better use of the time they spend in their cars.
Same thing, but different
As a German driver in the USA, you can basically hop into a rental car and drive off without risking an increased heart rate. But American traffic has more idiosyncrasies than you might notice at first glance. Here are a few pitfalls that autonomously driving vehicles in particular can still stumble into today.
The typical chrome-yellow school buses were introduced in the USA in 1939. When a school bus full of children stops at a bus stop and turns on its warning lights, all of the road users must also stop – including those in oncoming traffic. In other words, in the future the environment detection systems of our autonomous vehicles should be able to clearly identify whether the vehicle up ahead is a school bus rather than a public bus or a truck. During our drive we didn’t encounter a single school bus, because it was still vacation time. But our test drive engineer Matthias Kaiser defined numerous other “triggers” in the course of our drive. Whenever a traffic situation seemed relevant to the algorithm programming process, he set up a marker in the video of the test drive by clicking the buttons on the steering wheel.
“It’s important for us to know when the vehicle has to react,” Matthias Kaiser explained. “But it’s equally important to know when it doesn’t need to react.” While he was speaking, a motorcycle passed us on the right, simply whizzing between the lanes. In California, this is legal. Motorcycles are allowed to pass other vehicles at speeds of up to 30 mph (48 km/h) between two lanes on a highway. Our Mercedes S-Class didn’t even twitch – even though our camera sensors had recognized the motorcyclist much earlier than the driver, the front passenger, or the entire back row. After all, the S-Class didn’t need to know about the motorcycle, which was much faster than us and therefore posed no danger.
Overtaking is yet another peculiarity of driving in the USA. In addition to the motorcycles driving between the lanes, it’s also legal to overtake other vehicles on the right. Of course this is not a technical problem for our sensors, which feed the Active Lane Change Assist and other systems with information. However, it’s an additional “to do” for the programmers, who need to include this special regulation in the algorithm.
Botts’ Dots – a challenge for autonomous vehicles
Lane changing also involves another peculiarity that exists only in the USA: the “Botts’ Dots.” These are reflective raised dots on the roadway that pose a problem for automated cars, which “trip” over them if they have only learned to recognize white lane markers. They are used instead of the usual white lines on highways in many states in the USA. Automated systems such as the Active Steering Assist must be able to unequivocally recognize these dots in order to function reliably. But the “all clear” may soon be sounded in California. To prepare for the fully automated driving of the future, the state authorities are planning to gradually eliminate all of the approximately 20 million Botts’ Dots on California highways.
All lanes are not equal
In the USA, all lanes are not equal. In order to cope with the enormous volume of traffic from the suburbs, high-occupancy vehicle lanes ( HOV lanes, which are also known as carpool lanes) were created in the 1980s on multi-lane interstates and freeways. For the sensors and algorithms of automated vehicles, it is difficult to recognize them as special lanes and to distinguish them from normal lanes or exits. In addition, they are not always in the same place on a highway, and their markings are not uniform. They can be separated from the other lanes by two solid yellow lines or by metal planks, or they can be marked with painted diamonds. Autonomous vehicles must be able to recognize the number of occupants in a car and thus know whether or not the car is entitled to use the carpool lane.
Center turn lanes
However, HOV lanes are not the only special lanes in US traffic. There are also left turn lanes, which in some cases are approached by traffic coming from opposite directions. The drivers have to use their common sense to turn left at the right place, leaving enough distance between themselves and other vehicles. This is yet another challenge for autonomous vehicles – and a new trigger on Matthias’ list.
If you’ve successfully managed all of your lane changes, you will sooner or later come to an intersection without traffic lights. Here the German regulation “right before left” is not valid. Instead, the rule is “first come, first serve.” Only if several road users reach their respective stop signs at the same time does the “right before left” rule apply. Registering that, and ensuring that the autonomous vehicle reacts appropriately, is another “to do” for our team of developers. This is necessary, mainly because in case of doubt a polite American will tend to wait rather than be the first one to hit the gas pedal.
“Eyes” everywhere – traffic signs in text form or typeface on the roadway
One thing I realized during our test drive through L.A. was that the human brain is actually quite flexible in its perception and interpretation of patterns, in spite of its inferiority to a machine in terms of the speed at which it records and processes stimuli. We initially have to get used to new information, but then we capture it quickly – for example, traffic signs that are not uniform or are displayed on the roadway. By contrast, the sensor systems of autonomous cars have to correlate this information painstakingly on the basis of a great deal of data. This too was put on our engineers’ list in the form of triggers during our test drive.
One more word about Technology
On the screen installed in the cockpit, the test driver sees image data that are simultaneous translations of what the sensors of the car are perceiving. For a passenger, this is quite impressive. The sensors detect stationary and moving objects and use vector arrows to indicate their direction and speed of movement. They even recognize Botts’ Dots and the pylons of construction site markers. It’s fascinating.
After the test drive, we head back to the brand-new office of MBRDNA in Long Beach. Here, journalists and visitors can attend presentations and go on tours of the workshops and the emission measurement stations.
I’m excited about the new office space and also by the atmosphere. As I look around me, I see a whole group of young people who are writing codes, algorithms, test reports, press releases or tweets and working together across their various disciplines as a matter of course. OK, it’s not a mission to Mars, but I’m still a bit awed, because this group, together with many hundred other experienced colleagues in Sindelfingen, Sunnyvale, and Bangalore, will ensure that at some time in the future we will be able to travel autonomously, comfortably, and safely from A to B. Our day of test driving is over now. The test car and its crew have a short night ahead of them. They’ll have to get up at four in the morning to photograph the vehicle together with the engineers in the sunrise before continuing on to the final stage of the test drive at the CES in Las Vegas.