Drive it like you stole it.
Grand Theft Auto is being used to teach the artificial intelligence that controls self-driving cars.
Researchers are using open worlds built by Rockstar and other developers to replicate real-world data.
Using games like Grand Theft Auto is cheaper and less time-consuming than creating bespoke software for machine learning, according to a Technology Review report.
Researchers at Intel Labs and Darmstadt University have created software that can help AI differentiate between vehicles, pedestrians and other scenery in games like GTA. Doing the same with real-world data would take “thousands of hours” to collect and label the imagery.
“With artificial environments we can effortlessly gather precisely annotated data at a larger scale with a considerable amount of variation in lighting and climate settings,” said Alireza Shafaei, a student and the University of British Columbia.
“We showed that this synthetic data is almost as good, or sometimes even better, than using real data for training.”
Would you be happy in a car that learnt how to navigate the streets of Los Santos and the Cunning Stunts courses?