In the past, some have argued that games such as Rockstar’s Grand Theft Auto (GTA) franchise have promoted violence. However there have been instances where it has been found that it actually yielded some positive results, and it seems that over in Germany, researchers from Darmstadt University and Intel Labs have actually used the game to help teach self-driving cars how to navigate better in real life.
According to the researchers, while machine learning might be a preferred way of teaching computers do certain things, the amount of data required is immense, and sometimes data can be hard to come by, which is why the researchers have turned to video games like GTA in which information such as roads and obstacles and objects can be used to substitute their real life versions. In some cases they claim that data from games is “almost as good, or sometimes even better”.
What the researchers have done is create a software that can identify objects within GTA’s world, such as different objects and roads, and based on that they can use that information to teach self-driving cars the difference between pedestrians and other objects while driving. As to whether or not this method of teaching cars how to drive themselves will prove to be the most effective/efficient, it is a rather interesting approach.