Researchers Are Using "Grand Theft Auto" Video Game Technology to Improve Self-Driving Cars

Posted by: Jesse Kleib on 11/16/2021

Whelp, robots are officially teaching one another thing.

Ok, I have to admit that we’re a little bit late to the party on this one. Surely, there are some who never caught wind of this incredible story that originated back in 2016, though. Let’s just say that there are some pretty cool things happening with technology that are sure to have taken the upward exponential curve at this point.

Back in 2016, a German university by the name of Technische Universität Darmstadt decided to get creative to make the most of the tools available. The problem they faced was that properly programming an autonomous car would take thousands upon thousands of trips. These days, Tesla is gathering data constantly as drivers make their way through public streets. Computers are analyzing the movements that the drivers make in order to attempt to make these self-driving capabilities is even better.

Even with all of the data collected by Tesla, the system still isn’t perfect.

This is where some sort of simulation might help out the cars just a little bit. As it turns out, the German university decided to make the most of the video game, Grand Theft Auto 5. Researchers found that the game’s artificial intelligence is so realistic that they could actually use it to create meaningful data.

This data could then be used to teach self-driving cars how to be better on the road. Essentially, the university developed software that runs with the video game in order to help machine learning. They say that the game’s realistic appearance along with realistic and variable light and traffic conditions helped out the AI’s learning process.

By following along with the video below from DidYouKnowGaming?, we take a little bit deeper dive into how exactly the specifics work out. For those who think that robots are going to take over the world, they had better shield their eyes. There is a creepy yet exciting aura around watching robots learn by themselves.

0 comments