When the car drives by itself

by Ann-Britt Rognes and Jeppe Songe-Møller


Car on road

A few years ago, we looked with wonder at what AI could do for us, how AI could simplify everyday life and how self-driving cars were very exotic. A car driving itself belonged to the movies and the imagination. But now, a self-driving car is close to being a reality. All indications are that fully autonomous cars – powered by AI technology – will become a reality within the next decade. Soon you will be able to get into your car, close your eyes whilst driving at 110 km/h and trust that your car, and the computer controlling it, will get you safely to your destination.

1. AI development in the automotive industry

The automotive industry has invested heavily in AI and machine learning to achieve self-driving cars. One of the first steps was "Cruise Control", which allows a car to travel at a specific speed that the driver sets on the steering wheel. The car maintains speed without the use of the pedals. This was followed by "Adaptive Cruise Control". This feature built on the previous function, but now responds if the car starts to catch up (too closely) with the car in front by adjusting the speed. In cars manufactured after 2020, it has become common to have so-called "lane holders" that allow the car to stay within the drawn lines on the road.

One of the latest developments is "Intelligent Speed Assistance" (ISA). A camera in the car reads the traffic signs and the cruise control ensures that the car automatically adapts to the correct speed. The EU has decided that all new cars should have this technology, and all new cars (made in the EU) will have such equipment fitted from 1 January 2022. Estimates show that ISA can prevent 20 percent of the approximately 25,000 traffic deaths in Europe each year. I.e., about 5,000 lives could be saved. In addition to a speed barrier, all new cars will also automatically brake when encountering cyclists and pedestrians and a seat belt warning shall be included for all seats as standard.

All of these new technologies mean that we are ever closer to a self-driving car.

2. Degrees of self-driving cars

AI is gradually taking over more and more of the driver's functions of the driving, and the driver's role is becoming less and less important. The international standardisation agency, SAE, has defined five degrees of autonomy for vehicles:

Level 0: Automatic systems provide alerts but with no degree of vehicle control.

Level 1: ("Hands on"): The driver and automatic systems share control of the car. Examples include Adaptive Cruise Control, where the driver controls the steering and the car controls the speed, the parking assist, where the vehicle controls the turning of the steering wheel and the driver controls the speed. The driver must be ready at all times to take control of all functions.

Level 2: ("Hands off"): The automated system can take full control of the driving functions - throttle, brake and steering. The driver must be in control of the driving and ready to intervene immediately if systems fail.

Level 3: ("Eyes off"): The driver can safely distract from driving. The vehicle will respond to situations that require immediate action, such as emergency braking. Nevertheless, the driver must be prepared to intervene within a specified time, specified by the vehicle manufacturer, when the vehicle so notifies.

Level 4: ("Mind off"): This is like Level 3, except that the driver's readiness is not a requirement. I.e., the driver can sleep or leave the driver's seat. Such autonomous driving can only take place within defined areas/geographical boundaries, or in special situations such as congestion and traffic jams. Outside of these areas or situations, the vehicle must be able to stop driving and park safely if the driver does not take control.

Level 5: ("Wheel optional"): The steering wheel may be redundant. No human intervention required. Example: A robotic taxi.

3. Advantages of AI and cars

In recent years, AI has revolutionised the automotive industry. AI is used to make vehicles smarter and safer. The algorithms can analyse data from a car's sensors and cameras to detect obstacles, such as pedestrians, and make decisions about how to avoid them. AI is also used to create virtual maps that can help the car anticipate road conditions and react accordingly.

The potential benefits of AI-enabled autonomous vehicles are enormous. Autonomous vehicles are expected to reduce traffic congestion and increase road safety. AI can also help improve fuel efficiency, as cars can be programmed to run more efficiently.

AI will also help more people keep their independence for longer. Elderly drivers will, with the new technology that AI brings, be able to keep their right to drive for longer without compromising safety. Conditions that have previously led to elderly people being deprived of the right to drive will, to a greater extent, be remedied by the safety systems that AI brings into the cars. This in turn means that the elderly will keep their independence for longer, which will contribute to improving the quality of life.

4. Disadvantages of AI and cars

Of course, there are also potential dangers associated with autonomous vehicles. AI algorithms are designed to make decisions quickly and accurately, but they may not always be able to predict the unpredictable behaviour of people. In addition, hackers could potentially exploit AI-enabled vehicles for malicious purposes.

Self-driving cars also raise a number of ethical issues. In particular, a question will arise where autonomous vehicles are involved in accidents. Who then is liable? In manual driving, the driver of a vehicle can be held liable for any injury or damage that occurs. However, this may not be the case when an autonomous vehicle is involved, since the role of AI technology in the accident must be taken into account. Without a human driver, who should be held liable for any damages?

Ethically speaking, we will also need to consider how much we should trust AI. E.g., by allowing elderly persons to keep their right to drive for significantly longer and in cases where they had previously lost their right to drive, who will then decide where to draw the line? How should the line be drawn?

Furthermore, the use of AI in autonomous vehicles raises the issue of ethical decision-making. In the event of a collision, autonomous vehicles must make split-second decisions about how to react to avoid serious injury or death. Who determines the ethical considerations that play a role in these decisions? How do we ensure that the AI system makes decisions based on ethical criteria rather than convenience or cost-saving?

The use of AI technology, combined with the ability to track vehicle movements, may also lead to a loss of privacy. Who is responsible for ensuring that the data collected by autonomous vehicles is not used for any improper purpose? And will we be tracked everywhere and in everything we do?

5. The future

There is no doubt that AI technology has the potential to revolutionise the autonomous vehicle industry. As we open up more to AI technology in the automotive industry, and exploit the opportunities that AI provides, autonomous vehicles will become safer, more efficient and more cost-effective. AI is becoming an increasingly important part of the development of the car of the future. However, it will probably take some time before we are able to get into a vehicle that takes us from A to B without us having to do anything. The fully autonomous car is probably a long way away, but it is coming.

For most of us, the thought of a fully autonomous car is also something frightening, and before this becomes a reality, there is a lot that needs to be in place. Firstly, the population must be accustomed to the development of AI cars. Secondly, road networks and, not least, appropriate legislation must be in place.

In connection with regulation, there are many issues that must be taken into account and investigated. Who is to blame in the event of accidents, and can the human driver be punished? Where should the line be drawn between human and software in respect to liability? Privacy considerations will also come into play. As the car registers more and more information, and often also takes photos and films, are we okay with cars collecting our personal data, and are we comfortable with our personal electronic tracks being used for AI development? Increased use of AI in cars will push legal boundaries, and privacy and security considerations will be essential topics. We will need to be careful that the traces we leave behind do not fall into the wrong hands and are not used in the wrong way.

Do you have any questions?