Researchers are using the vibrancy of Grand Theft Auto 5’s world to help train driverless cars in dealing with unpredictable scenarios quickly and safely.
The use of virtual worlds helps developers to optimise their AIs within a safe environment before rolling out the technology into the real world. Players of GTA 5 will be aware the game offers various densities of traffic and pedestrians which help to simulate real world scenarios. In cities, the volume of traffic on the roads and footfall on the sidewalk are typically high and the pace is quite slow. In rural areas or the highway, traffic is typically reduced but faster moving.
“GTA 5 is the richest virtual environment that we could extract data from,” said Alain Kornhauser, Princeton University Professor of Operations Research and Financial Engineering. Considering the games has 262 types of vehicles, over 1,000 various pedestrians and animals, 14 different weather conditions, and a large number of bridges, traffic signals, tunnels, and intersections, it’s little surprise the game offers a rich environment for researchers.
“Just relying on data from the roads is not practical,” said Davide Bacchet, who leads autonomous electric car startup, Nio. “With simulation, you can run the same scenario over and over again for infinite times, then test it again.”
While many players seek out wanton destruction in GTA 5 – whether causing chain reaction explosions or knocking over pedestrians – these are the exact outcomes the researchers will be looking to prevent.
Most car-related incidents are due to human error such as tiredness, distraction, or being under the influence of alcohol or drugs. Current accidents involving self-driving cars are often found to be caused by another human-controlled vehicle. Over time, it’s expected self-driving cars will significantly improve road safety.
To appease road safety bodies and get the use of autonomous vehicles approved; it’s likely going to take proof of billions of miles of virtual and real-world tests.
What are your thoughts on using games for training driverless car AI? Let us know in the comments.
Interested in hearing industry leaders discuss subjects like this and sharing their IoT use-cases? Attend the IoT Tech Expo World Series events with upcoming shows in Silicon Valley, London and Amsterdam to learn more.