While nothing can replace actual road time for autonomous vehicle testing, the popular video game, “Grand Theft Auto,” is helping engineers and scientists understand and simulate road time for their programs. Engineers are able to re-create edge case scenarios that may be difficult to test in real life to assessthe intelligence and reaction of the vehicles in design. The realistic scenery of the game along with realistic road design may help the artificial intelligence recognize patterns and elements found in city streets. The goal is to train the AI used on board autonomous vehicles to respond to events on the road correctly through machine learning and massive quantities of experimentation. That’s where using a computer simulation is so handy.
Simulated Driving, Real Data
The algorithms used to pilot autonomous vehicles are able to log millions of miles driven without the real-world costs of driving i.e. gas, battery charging, maintenance etc. Besides reducing the costs associated with the development of these large projects, it may end up saving lives as part of the learning process can take place in a virtual reality. Additionally, some researchers believe the “synthetic data” is better than real-world imagery for machine learning. At the end of the day, it is highly unlikely that your vehicle will attempt to assault somebody because it learned to drive in a Grand Theft Auto videogame. With innovative technology advancements like this, we may not be so far away for mass public use of autonomous vehicles.