Masters of Smart

Cars are getting smarter. We no longer have to park ourselves and automatically stay within the lines... But the question is whether we can ever really get behind the wheel. In order to function safely on all roads, under all circumstances, there appears to be a lot of work to be done.
Tesla Cybertruck

Text: Eltjo Nieuwenhuis

Image header: Tesla Cybertruck

In 1968, the Volkswagen Type 3 was released with computer-controlled fuel injection. This was the first computer ever in a production car. At that time, computers were still the size of a refrigerator and could do less than, for example, a smartphone. So it was quite an achievement that VW got their electronic brain small enough for use in a car. The feat turned out to be the harbinger of a digital revolution in the automotive industry.

The same year came Stanley Kubrick's 2001: A Space Odyssey out. In this classic film, a talking AI supercomputer (HAL 9000) takes over a space station and then murders the crew. With HAL, the film gave shape to the fear that already existed at the time for computers and artificial intelligence. This fear of an out-of-control digital brain obviously still exists, also in the automotive industry.For some potential car buyers, their fear of cars being too smart is a real factor in their decision-making process. Partly for this reason, manufacturers have often been reluctant to advertise how computer-controlled their cars actually are. This is mainly motivated by marketing departments that prefer to remain silent about things that customers fear. Recently, however, marketers have increasingly started to 'admit' that there is AI in their cars. It has even become a selling point. Artificial Intelligence is becoming less scary. We are becoming more and more familiar with the everyday applications of AI, so the mythical is disappearing. We now find it quite useful when cars help us a little with options such as Lane Assist, Park Assist and Traffic Jam Assist. Motorists appear increasingly willing to hand over some autonomy.

VW Type 3, the first car with a computer on board

Level 0 to 5

Our cars have gradually taken over more and more tasks from us. In the early twentieth century, you still had to crank them by hand, until Cadillac introduced the starter motor in 1912. The first automatic choke appeared in the XNUMXs, so drivers no longer had to adjust the air/gasoline mixture with the hand choke. Of course, these are small steps compared to cars taking over the controls for us. However, the automotive industry is definitely working towards autonomous driving.

In 2025, cars with Level 5 autonomy will be allowed in the UK. The different levels of autonomy start at Level 0, the 'driver' level, where there is actually no computer assistance. Level 1 is also known as 'feet off'. An example is adaptive cruise control, where the car maintains a constant speed, but brakes if you threaten to get too close to the vehicle in front. At Level 2, or 'hands off', the car can steer, brake and accelerate, but the driver remains responsible. So even if you are not actively driving, as a driver you have to keep your hands on the wheel to intervene.Level 3 is called 'eyes off', which means that the car drives independently in uncomplicated situations, such as on the highway. Is the highway that simple? In fact, yes, because there are no side roads, traffic lights, cyclists, playing children, etc. However, if the situation becomes more complicated, the car warns that your full attention is required. Level 4 is called 'mind off' and means that the car drives independently, but the person in the driver's seat can intervene. In bad weather, for example, the cameras, sensors and laser measuring systems work less well, and the driver still has to go to work. And then there is Level 5, the 'passenger' level. Every occupant is a passenger and there are no steering wheel or pedals in the car. And Level 5 will be permitted on British roads from 2025.

Full autonomy

In the Netherlands, Level 2 is currently the highest permitted level of autonomy. Level 2 is permitted almost worldwide and already goes quite far. Tesla, for example, has been handy with this. Many of their models are actually Level 4, but because the supplied manual states that you must always keep your hands on the wheel and eyes on the road, they are formally only Level 2. Yet in the US, for example, they are often used as Level 4. YouTube is full of videos of drivers sleeping on the highway, even in states where it is illegal. Levels 4 and 5 are now permitted in more and more American states. It is no coincidence that countries such as China, Germany and the US allow far-reaching autonomy on their roads. This is how the cars of the future are tested and the AI ​​can learn. This is all part of the race to bring the first, best, and safest self-driving car to market. Artificial Intelligence plays a crucial role in this. AI differs from 'normal' computers and algorithms in that it can constantly learn something new. The self-learning computer starts with instructions that are as complete as possible, but can always encounter new situations. By means of machine learning the computer analyzes these situations and makes small adjustments to the instructions in order to constantly improve them.

So from 2025 we'll be zipping around in driverless cars?

With everything going on, you might think that full autonomy is just around the corner. Because Level 5 autonomy will soon be allowed on British roads, YouTube is full of successful tests, AI has taken off, manufacturers have been showing beautiful concept cars for years. The signs seem good, but the reality is more complicated. Even though, for example, self-driving shuttles already function on fixed routes under controlled conditions, this says little about complete autonomy. To be able to function safely on all roads, under all conditions, there is still a lot of work to be done. And some manufacturers are already warning about this. BYD (Build Your Dreams) is the largest Chinese manufacturer of electric vehicles and threw a bucket of water on the autonomous dream earlier this year. A BYD spokesperson said at the Shanghai Auto Show: “Fully autonomous self-driving is still very, very far away, and basically impossible.” Of course, such a statement to the press is full of politics and marketing, but that does not make it unfounded. Because we are not there yet. It is now possible to drive safely autonomously on the highway, on known routes or with installed beacons. But it becomes a lot less safe if you drive on an unknown route between hectic traffic, in the snow. Saying that this will never be possible autonomously is like saying sixty years ago that a return trip to the moon is impossible. But it is extremely difficult.

Bentley EXP 100GT

Real prediction

About five to ten years ago, many extravagant self-driving concept cars were released, but currently the manufacturers are more modest and they mainly test and learn. Various manufacturers who are working full-speed on autonomy warn that Level 5 may be permitted and applied here and there, but may only be rolled out commercially after 2035. But let's listen to Hyundai. The South Korean company has the world's longest running fleet of self-driving taxis operating in Las Vegas. Their IONIQ 5 Robotaxis have already provided more than 100.000 rides and caused zero accidents. And Hyundai predicts that from 2030, cars will be able to drive themselves. Which of course does not mean that the public can immediately buy them, so that will take you a long way towards 2035.

In 2021, during the Tokyo Olympic and Paralympic Games, self-driving shuttles drove people from A to B. These e-palette shuttles even had supervisors/drivers present, but the project mainly became a demonstration of the need for more testing. Neither shuttle nor driver could prevent the vehicle and a blind Paralympian from colliding. The human and artificial intelligence together turned out not to be 'smart' enough to prevent an injury. And while we often overlook human errors, AI and self-driving must meet stricter standards. Rightly so, because technological progress must provide improvements. That is precisely why this process will take a number of years. Cars have become very smart in a short time, but we are not yet at Level 5.

Hyuandai Le Fil Rouge