From the driving assistance to the autonomous car : 5 levels of artificial intelligence integration

Artificial intelligence existed since the late 1950s but has only really been used for about ten years. The main reason for this late development is that today we have a large amount of data necessary for the proper functioning and development of artificial intelligence.

The term artificial intelligence was coined by John McCarthy in 1955. This takes place at the Dartmouth conference of about ten scientists. The objective of this giant brainstorming is to prove that it is possible to reproduce human intelligence and reasoning in a machine. Two types of artificial intelligence are then distinguished. Strong artificial intelligence that is able to reproduce intelligent behaviour and reasoning but also to create a self-awareness for the machine, and thus make it capable of “understanding” what it does. Thus, a strong AI can analyze, reason, be rational. On the contrary, a weak artificial intelligence does not understand what it does. It reproduces as faithfully as possible a process whose outcome can be known in advance, and leaves no room for improvisation.

The automation of a vehicle consists in taking back the perception and actions of a human driver, as well as the 5 senses a person use. The vehicle must be able to understand the environment around it, to analyse and react accordingly. To achieve this, the vehicle must be equipped with sensors, central computing power and electronic controls capable of reproducing human behaviour. In addition to analytical skills, communication between vehicles must be added to provide maximum real-time information on the overall perception of the vehicle and the environment in which it operates, so that potentially dangerous situations can be anticipated. One of the undeniable advantages of automated vehicles is the reduction of accidents and thus road mortality, since 80% of serious road accidents are due to human error, the automated car could be a solution.

The 5 levels of automotive automation

Artificial intelligence can be integrated at 5 levels in the use of a car.

Assisted driving

The first level is assisted driving. Driving can be assisted, which makes it safer and more comfortable, with reversing radars, cruise control that adapts to the distance of the vehicle in front or the automatic braking function in town to avoid accidents. Assisted driving is already well established in vehicles. Nevertheless, the use of assisted driving functions does not take control of the vehicle, the driver is still a major actor in his driving and cannot take his eyes off the road. This level of use of artificial intelligence is quite common, and has been gradually implemented since the 1990s.

Semi-automatic driving

At this level the driver is still responsible for driving and must remain focused on the road. Semi-automatic driving offers steering and guidance assistance during traffic jams as well as automatic brakeing/acceleration or parking. An executive assistant works as follows, combining level 1 technologies and adding automated longitudinal and lateral control. This allows micro-corrections to be made on the steering wheel to keep the car in line. However, if the driver releases the steering wheel too long, an alarm sounds, then after a while the vehicle brakes by accents, if the driver does not react, the car will gradually brake until the vehicle stops completely. Semi-automatic driving appeared in the 2000s on top-of-the-range sedans and is becoming more and more popular.

Highly automated driving

Level 3 allows the vehicle to take over part of the driving. This allows the driver to let go of the steering wheel in certain traffic situations, such as motorways, traffic jams and car parks. He must be able to resume driving quickly, especially when work is being done. There are already cars capable of performing these functions, but the regulations do not allow this type of driving.

Fully automated driving

Fully automated driving is level 4, also known as emancipation. It is an advanced level of automation that allows the vehicle to handle complex urban traffic situations. However, the driver must be able to take control of the vehicle if necessary. Unlike the previous level, the driver can take a nap and in case of a dangerous situation or requiring the driver’s intervention, a bell sounds. If the driver does not react, the system will put the car in a safe place by stopping it. Moreover, from automation level 4 onwards, it becomes more than necessary to have a very detailed map to know the exact position of the vehicle on the road. This type of vehicle is still only in the prototype stage and will not be marketed until 2020 for the most optimistic manufacturer and developer.

Autonomous cars

Autonomous cars are the latest level of automation. At this level, the car does not need a driver. It is not necessary to have the ability or licence to drive and therefore no steering wheel or pedals. All driving functions are supported by the vehicle system. This new level of automation offers disabled people new mobility solutions. Autonomous cars are technically very complex and demanding to design. As a result, this type of vehicle will not be deployed until 2025/2030 and will be reserved for low speed areas in urban traffic.

Automated type 4 or 5 cars are under development, but will not be democratized quickly. The cost of these technologies is very high. Above all, the regulations are also not ready to accommodate level 4 or 5 vehicles. Advanced vehicle automation raises liability issues in the event of an accident, for example, which must be resolved by the relevant stakeholders before this type of product can be democratized. Indeed, in Europe, the Vienna Convention imposes on drivers the responsibility of controlling the vehicle, which is why vehicles, whatever their degree of automation, are equipped with sound reminders. There are still many answers to be found in the development of autonomous vehicles, but we can say that artificial intelligence makes driving safer and easier.

If you liked this article, you will be interested How is AI and AR merging in the future ? and the webinar How to use Google Ads and AR to increase your revenue for retail and ecommerce ?