The Grand Theft Auto series has been criticized for its 'morality' scale and prostitutes, but it's finally getting put in the spotlight for something good: helping teach autonomous vehicles. That's right, driverless cars are learning from GTA.
If you've played GTA, you've probably jumped in a car and drove around town, probably crashing it into various things and people. There's one thing that happens no matter what you crash into – there's a reaction from other drivers/AI. These reactions and the environment are actually pretty good.
As it turns out that GTA is one of the simulators that are being used to drill driverless cars and help teach them how to respond in case of an unplanned event on the road.
It all started last year when scientists from Darmstadt University of Technology in Germany and Intel Labs discovered a method that pulled visual information from Grand Theft Auto V. After developing a way to pull the visual information, researchers are "deriving algorithms from GTAV software that’s been tweaked for use in the burgeoning self-driving sector."
Why GTA though? According to one professor, it's because GTA “is the richest virtual environment that we could extract data from.” The Professor, Alain Kornhauser, the Princeton University professor of operations research and financial engineering who advises the Princeton Autonomous Vehicle Engineering team, isn't wrong.
GTA is equipped with 262 types of vehicles, and plenty of pedestrians and animals that will do almost anything – run out in the road frantically, constantly stand in the way and more. On top of that, there are 14 weather conditions, plenty of bridges, traffic signals, tunnels, and intersections. It's pretty much real-life minus the driverless cars running off the road and hitting real people.
It's the perfect training ground.
