For a self-driving car to really live up to its customers' expectation, a grueling three million miles of virtual road test is needed to test its latest development, at least, according to Google's self-driving car team. According to CNET, while Google believes that actual road test is still the best option when testing a car, they have also subjected the project to a grueling three-million mile virtual road test every time they add or enhance an important car feature.

Based on the January report of Google's Self-Driving Car Project, the team listed an average of 10,000 to 15,000 miles of public road test per week. By integrating a simulator on their project, Google said that they don't have to put up with releasing a unit from the garage every time, but rather use the simulator to perform initial tests.

Google also believes that by using a computer to test their cars, they are able to record every single point of the testing, enabling them to accurately recall information that plays an important role in the development of the car.

In the report, the team said, "Our engineers are regularly adding new capabilities and refining the car's driving by making improvements to the software. Before we roll out any change to our fleet of cars, we first test it extensively in this virtual environment."

Citing the difference on how traffic lights are installed and profiled in California and Austin, Texas, it allowed Google to properly calibrate the car's ability to detect traffic lights. In Mountain View, California, traffic lights were installed vertically, while in Austin, Texas, they were significantly adjusted to a horizontal and dimmer profile. Being able to collect those simple yet equally important information allowed Google to update its database in more detail.

CNET also said that Google's use of simulator also allowed the car to determine the next move when the driver suddenly decides to take control of the car instead of letting it drive alone. In the report, the Google team said, "Our simulator can also help our test drivers and engineers quickly identify any areas of improvement. Each time a test driver takes over from the self driving car, we're able to play back the exact situation and predict via simulation what could have happened if the car had been left to drive itself."

Google did not downplay the importance of real-world testing, citing that it is equally important as the simulator.