Society is on a precipice of a future with self-driving cars and one A.I. program could very well help to get us there.
For decades technology has been advancing by rates that have outpaced many individuals’ abilities to comprehend and keep up with. Artificial Intelligence or A.I., is one sector in particular whose improvements never fail to continuously astound. Perhaps even indicating we are on the precipice of a Skynet future? Let’s hope not! Still, regardless of what the future holds, A.I. at present is still exceedingly impressive. In fact, NPR reported that one type of A.I. program, in particular, has perked the ears of a Stanford Professor who thinks its real-world applicability could help with self-driving car technologies.
The A.I. program in question is called GT Sophy, which is short for Gran Turismo Sophy. Gran Turismo is a little racing game you may have heard of developed by the brilliant minds over at Sony. The reason why the A.I. program is called GT Sohpy is because of its ability to best some of the top Gran Turismo racers in the world.
GT Sophy team leader Peter Wurman explained that the A.I. was able to get that good at the game all on its own. Wurman said his team was able to facilitate this by hooking up the A.I. program to approximately 1,000 Playstation 4 consoles that were constantly throwing simulations at it. Through the simulations, the A.I. figured out the best ways to race and in a matter of 10 days, it could beat the majority of players in the world. It proved it when it singlehandedly won a huge e-sports race in Tokyo.
While this is all very impressive, you might be asking yourself what this has to do with autonomous vehicles? This is where Standford’s mechanical engineering professor Chris Gerdes’ thoughts come in. Gerdes thinks that the technology that made the GT Sophy A.I. program so successful could be adapted to be utilized in the development of self-driving cars.
Gerdes, who in addition to being an esteemed professor races cars himself, relayed that he spent quite some time observing GT Sophy’s applications and abilities. He concluded that some of the A.I. program’s actions while playing Gran Turismo were akin to intelligent decisions that a human driver might employ. “It turns out that Sophy actually is doing things that race car drivers would consider to be very intelligent, making maneuvers that it would take a human race car driver a career to be able to pull some off…,” said Gerdes.
What Gerdes is implying with his conclusion is that A.I. programs like GT Sophy, if integrated into real-world autonomous cars, could make decisions while on the road comparable to, if not better than, actual human drivers. Over time, an application such as that could go a long way to bolster the confidence of skeptics unwilling to test the unknown waters of self-driving car technology. However, despite the potentiality that an A.I. program like GT Sophy exudes, these early observations are simply that – early observations. Gerdes highlighted that “It’s not as if you can simply take the results of this paper and say, ‘Great, I’m going to try it on an autonomous vehicle tomorrow.” Still, he remained adamant that at the very least GT Sophy could help individuals to “…keep an open mind about the extent of possibilities here with A.I. and neural networks.”