Cars of the future will be driverless: they will design a route and safely ride paying attention to traffic lights, road signs, and traffic. All the abovementioned will be possible due to artificial intelligence and a complex system of sensors and cameras. We currently cannot witness driverless cars on the city roads but their prototypes are going through the stage of development and tests.

What is a smart car?

How does artificial intelligence work for a car? This is a tool performing some function instead of a human. As explained by Head of AI and Research at Yandex Mikhail Bilenko, artificial intelligence is aimed to process input data and further solutions for cars. According to him, cruise control is also an example of artificial intelligence, though primitive.

Another application case is a voice assistant built in a board computer of modern cars. It recognizes driver’s speech, provides weather forecast, browses the Internet, and switches on music.

However, these are only first steps. A car can be called autonomous only on condition that everything is under control of artificial intelligence.

AI Conference: How do driverless cars see and learn? AI for transport 1

AI and its sensors

A central processing unit accumulates and collects data about car speed, performance of its systems, landscapes, other transport means, pedestrians, road signs, and traffic lights. The following devices are used to gather all the information:

  • cameras;
  • GPS;
  • light radars;
  • conventional radars;
  • movement and speed sensors.

Data is uploaded to a computer brain that is a multilevel network of electronic neurons. Then, a computer makes a decision based on its programmed algorithms. It moves with speed and priority set by rules, signs, traffic lights, and road surface marking; yields to pedestrians and keeps a safe distance with other cars.

Future created today

Driverless cars are today’s reality created and tested by various companies including Google and NVIDIA.

  • Google presented its self-driving car in 2010. It sees a road using two roof cameras.
  • AUDI TTS, or Audi Shelley, a driverless racing car that was presented in summer 2012. The engineers had to train artificial intelligence in order to design such a car.
  • Volvo released about a hundred self-driving cars to test all of them under real road conditions.
  • Nissan designed a driverless car based on an electric auto NissanLeaf.
  • General Motors presented a self-driving Cadillac SRX. Equipped with all kinds of sensors, the car underwent a test road in Michigan.

Prototypes of driverless cars have problems that are currently difficult to solve. They include a possibility to react to emergencies (for example, pedestrians appearing on the road out of sudden) and make decisions under rough conditions: rain and snow, potholes. To ensure proper functioning of a car, manufacturers should improve sensors and create more complex AI-fueled models capable of processing information several times faster.