Self-driving cars are the future. That much is clear. But to what extent are they self-driving? Should humans still intervene? And don't these cars make mistakes? We have listed the different levels of (self-driving) cars for you, together with the disadvantages, but certainly also the advantages of (more of) these cars in traffic.

A self-driving car often operates with multiple cameras that use AI to recognize objects and respond to them. This way, they can avoid other cars and also automatically stop for a red traffic light. Some car manufacturers use sensors, radar, and lidar in addition to the cameras. These cars often have an additional rotating sensor on top of the car.

Gradually, various levels of self-driving cars have emerged. Each level bringing you one step closer to fully autonomous cars, in which you as a human no longer have to do anything.

Six (actually five) levels of self-driving cars

Level 0 cars are currently by far the most common in the world. The newer cars have a camera or sensor to assist when reversing, but you have to brake yourself to prevent accidents.

Level 1 cars have some control over what the car itself does in a number of situations. This includes, for example, 'lane assist' to ensure that you stay within the lane lines and radar-based 'adaptive cruise control' that independently maintains sufficient distance between you and the vehicle in front of you. Emergency braking systems are also a part of level 1 cars, together with the blind spot monitor. Since 2022, all new cars in the EU must have certain level 1 functions. This is to drastically improve traffic safety.

Level 2 cars can drive completely independently in certain situations by steering and accelerating or braking themselves. Cars with a parking function are also level 2 cars. The driver must always keep his hands on the wheel and be able to intervene.

Level 3 cars can drive on a normal road, along a specific route entered by the driver. They don't have a major failsafe, so they still require human drivers to intervene. The car drives completely on its own, but if it cannot solve a certain situation, you as a human being have to intervene. This level of self-driving cars is often skipped, because the point of independent driving is outweighed by the fact that people still have to pay attention and be able to intervene at any moment.

Level 4 cars use human drivers as a last resort, but can do some more complicated things themselves. They can drive around a certain area independently or perform certain tasks completely independently. The car stops driving itself in a situation it cannot resolve until a human driver can take over. It usually does this by parking the car.

Level 5 cars require no human intervention at all and in many concept models it is not possible to intervene yourself. They often have no steering wheel. They can do everything a human can do, but without making human errors.

Limitations of (Partially) Autonomous Cars

Current self-driving cars often use cameras. However, this method is very sensitive to light and weather conditions. Rain or snow can impair the camera's performance, causing it to overlook important details. Tesla has stopped using radar and ultrasonic sensors, making their cars purely dependent on cameras. However, these cameras have a blind spot low in front of the car, causing important things not to be detected by the cameras. This can be dangerous if, for example, a dog or a small child is in front of the car. This is why other companies like Waymo also use radar and lidar, often in combination with sensors. This increases safety but also adds to the cost of the extra technology.

Accidents sometimes occur with self-driving cars that could often have been prevented by human drivers. Most of these are mistakes that humans do not easily make, but that are difficult to teach AI. Consider, for example, someone who crosses the road, but does not walk at a pedestrian crossing, or someone who is jaywalking. Cars also often do not 'understand' that people can disappear behind an object and then reappear. We as humans understand this and can therefore take action in time.

Autonomous cars can prevent many traffic fatalities.

Despite the fact that (largely) self-driving cars still cause accidents, they do cause a large decrease in the number of accidents. A self-driving car often responds faster than a human and that can save a large number of road deaths per year. It is estimated that it could save around 1.2 million deaths per year worldwide, if only self-driving cars were driving around. Of course this is a scenario for the future, but the more AI supports us in our driving, the safer traffic becomes. Let's embrace the new gadgets and use artificial intelligence to support us, even if we cannot or do not want to (completely) outsource driving.