In its ongoing effort to make sure that its self-driving cars will run as safely as possible, it looks like Google has its data centers simulate no less than 3 million miles of autonomous driving every day. That might seem like a lot of driving at a glance, but data shows that Americans actually drove 2.7 trillion miles in the year 2000 alone, which means that Google’s simulations are actually on point. Obviously, engineers want to use the data obtained from these simulations to perfect various algorithms that animate autonomous cars. Furthermore, the simulations also allow the company to create specific scenarios based on real-world events such as adjusting the speeds of cars during a highway merge.
Various improvements can then be designed and implemented back into the simulator in order to ensure that things will always go according to plan when the cars will actually hit the streets. However, before any code changes are actually applied, the simulator “re-drives” its complete driving history of more than 2 million miles. Even after the code goes live, Google still tests it with 10 to 15,000 miles of autonomous driving each week. It certainly appears that the company is taking things seriously when it comes to autonomous vehicles, but then again, that’s exactly what we should expect from a tech giant such as Google.
