see. think. act.



AI: Cars with "Gut Feelings"

Min Reading Time
Tags: AutonomousDriving, ArtificialIntelligence
Artificial intelligence is crucial in the development of autonomous driving functions since there are an infinite number of situations in road traffic that cannot be programmed in advance.
Arnold Schlegel, June 25, 2018
Arnold Schlegel is a ZF engineer and expert for autonomous driving and artificial intelligence.
An everyday situation in traffic: An experienced driver wants to turn into a roundabout. Another vehicle that has set its turning signal is approaching from the left, so the waiting driver can go ahead and enter the roundabout. However, something doesn’t seem right for the waiting driver. He stops and thus prevents a collision – because in fact the car in the roundabout simply continues driving despite its turn signal being set.

If you ask the waiting driver for the reasons for doing this, he will refer to his intuition. That’s how you describe the ability to draw conclusions from various information and to make use of your experience.
Without consciously thinking about it, the driver conclude from the speed, position of the wheels and the direction in which the other driver is looking that the car is not going to turn. From his experience he also knows that rules are sometimes not followed, that is, cars do not turn even though they have their turning signal set.
Normal software could not react like this, but this will be possible in the future with artificial intelligence.

While a normal program needs every context deterministically and only works in precisely known situations, algorithms are comfortable with abstraction. However, they need to train this ability just like the driver.

Autonomous driving in the starting blocks

Autonomous driving in the starting blocks

To many people, cars driving us autonomously from A to B on all roads and under all conditions continues to sound like a distant dream of the future. However, the necessary technology is already in the starting blocks: various automotive assistants are already using artificial intelligence. The following examples show this:
  • When it comes to traffic sign recognition, these machines are already superior to humans. Thanks to artificial intelligence, they have a success rate of 99.5 percent, and humans only reach 98.8 percent.
  • While well-known voice assistants in the car depend on clearly defined commands and input patterns, AI also understands the sentence “I’m cold” as a request to turn up the heating, for example. The system is constantly learning through updates after analyses of other drivers.
  • Smoothing over potholes: Cameras and sensors in the chassis can detect and compensate for road damage. In addition, other road users and the traffic control centre are informed, which can take care of repairs or ensure that the route planning of other vehicles avoids the hazardous location.

Artificial intelligence at the wheel

Artificial intelligence at the wheel

In order to be able to hand over control of the entire vehicle to the AI in the future, all these and many other systems must be consolidated in a central control unit. This requires one thing above all: sufficient computing power to evaluate the overwhelming amount of data from cameras, LIDAR, radar and other sensors in real time.
Experts define five autonomous driving levels, from level 0 (no automation) to level 5 (full automation).

This is made possible by high-performance control units such as the ZF ProAI developed by ZF, which handles up to 30 trillion operations per second (TOPS) with a power consumption of only 40 watts – and meets the strict standards for automotive applications. This was demonstrated by the central computer, for example, in a development vehicle with autonomous driving functions according to level 4 at the beginning of this year.

Dream Safety

Can a car stand still at a trade show in Las Vegas and simultaneously drive on a road in Friedrichshafen? The newly developed Level-4 autonomous "Dream Car" from ZF makes this possible.

Cars Learn to Have “Gut Feelings”

Cars Learn to Have “Gut Feelings”

One of the biggest challenges is the training of AI systems. How do you test a system that is designed to solve unforeseen problems? Virtual training and software-in-the-loop methods will make a contribution to overcoming this hurdle soon.
On this basis, autonomous vehicles could significantly reduce the number of accidents. Networking the autonomous cars could also result in an intelligent traffic guidance system with a previously unknown level of efficiency, which would ultimately succeed in preventing traffic jams. One day, for example, the artificial intelligence in the car might be able to check the position of the front wheels, the speed of vehicles and the direction of vision of the driver sitting in them – and let the driver who has his turning signal incorrectly set pass through the roundabout with aplomb.

Related Topics