How computers (and AI) changed the way we drive

The first car was patented in 1886 by Mr Karl Benz. The car was a simple tricycle with a little engine which had a maximum speed of 16 km/h. From that point on, cars developed in size and power until they become advanced enough to substitute horses. Nevertheless, at first, only the rich could afford them, the cars were all custom made for each client and cost an absurd amount of money. The advent of cars for the masses first occurred in the U.S. in 1913 when Henry Ford started mass-producing cars his prototype, the Ford Model T, was designed to be produced in bulks to reduce costs and labour and maximize efficiency. Europe had to wait a bit more as most of the continent only had mass-produced cars after World War II. The same occurred in Japan and the rest of Asia, while in other states, like India, cars started being popular only in the late 1960s.

The Benz Motorwagen. The first car ever.
The Ford Model T production line

The technological advancement of cars never stopped during this period, and the focus changed, cars were affordable, so the aim was to make them as safe as possible. Lights were first put on cars to be able to see and be seen. Mirrors were placed to be able to look behind without too much distraction, and seatbelts were placed to save lives.

The first time a computer was placed in a car for security purposes was in 1971 when Chrysler, a U.S. car company, first installed the Assisted Braking System (ABS) in one of its models, the Imperial. This computer used sensors and invasive control to avoid skidding while braking, since skidding (sliding) massively reduces traction. Drivers did not have to measure their braking in order to be able to brake safely; they could stomp on the brakes and let the computer do the rest. This also introduced braking while turning -which before could only be done by experienced drivers that could control the braking by themselves- since in case of skidding the car would go straight.

The purpose of the ABS

During the 1970s airbags were also introduced but drivers had to wait until the late 1980s for them to be fully reliable. During a crash, sensors would send information to the car’s computer -the Electronic Controller Unit (ECU)- including the angle of impact, type of collision and severity of impact. Using this information, the ECU’s algorithm determines whether the collision requires airbag deployment or not.

An ECU, the brain of modern cars

The 1908s also saw the introduction of the ESC (Electronic Stability Control) which continually monitors car handling while driving. In case the ECU detects a loss of control, the ESC applies braking to the wheels (even individually) or reduces power permitting the driver to regain control. It is especially useful when raining or snowing, situations in which car handling is massively reduced due to traction loss.

How the ESC can avoid an accident

Nowadays all of the previously mentioned systems are obligatory in all production cars in states such as Canada, U.S., the European Union, Japan and others.

The new era of security relies more on cameras (those substitute sensors) and Artificial Intelligence (that substitutes regular computers). Some modern cars are equipped with partial self-driving systems such as the ones seen on Teslas. Partial self-driving means that the ECU can assist in controlling speed and steering, but the driver must have hands on the steering wheel and be ready to take over. Nevertheless, most new cars come with some sort of crash avoidance system (CAS) or autonomous emergency braking (AEB). These systems, which the EU plans to make mandatory in the near future, override the driver’s input in case the ECU’s AI detects that the driver’s inputs are no longer safe for the driver itself (going too fast while facing an obstacle) or others (going too fast while a pedestrian is in front).

Cars are becoming more and more autonomous, out of the conventionally accepted five types of autonomous vehicles; a few cars on the market have already reached level 3. In the next few decades, technological development is expected to reach level 5. The major obstacles for driverless cars (level 5) are state regulations since technology moves much faster than law-making.

How the AI of a Tesla Model S sees the road.

It is safe to say that our generation will experience the transition from human-controlled cars to driverless cars. The cost of AI technology will drop as development continues making autonomous cars more affordable for the public, and soon enough it will be weird not to see a car drive itself on the highway.

Sources

Williams, K. (1994). Cars: Analysis, history, cases. Berghahn Books.

Lugano, G. (2017, May). Virtual assistants and self-driving cars. In 2017 15th International Conference on ITS Telecommunications (ITST) (pp. 1-5). IEEE.