Google’s self-driving cars are safer than human drivers

Macworld Australia Staff
8 August, 2012
View more articles fromthe author

Google’s self-driving cars are safer than actual drivers, with the company announcing its new vehicles have completed 300,000 miles (482, 803km) of test-drives in a ‘wide range of conditions,’ without any reported accidents – an improvement on incidents recorded in previous testing with drivers behind the wheel.

Mashable Tech have crunched the numbers, to compare Google’s latest series of tests with statistics on drivers in the US.

“The average U.S. driver has one accident roughly every 165,000 miles”, Mashable Tech writes.

“Here’s how we got that figure: our average mileage per year is 16,550, according to the Federal Highway Administration; the average length of time we go between traffic accidents is 10 years, according to Allstate. (In particularly safe cities such as Fort Collins, Colo., that number can rise to 14 years — which is still no match for Google’s 300,000 miles.)

The Google self-driving project uses Toyota Priuses fitted out with cameras, radar sensors and laser range-finders to detect and monitor traffic, while advanced software accesses Google Maps to navigate the roads. In previous testing Google had used two human drivers in the cars, in case they were needed to take over. Now the company says it will use only one driver per car moving forward.
But, Google admits it still has a long way to go before its self-driving cars are deemed roadworthy.

“To provide the best experience we can, we’ll need to master snow-covered roadways, interpret temporary construction signals and handle other tricky situations that many drivers encounter,” writes Google’s Engineering Lead overseeing the project in a blog post.

“For now, our team members will remain in the driver’s seats and will take back control if needed.”


One Comment

One person was compelled to have their say. We encourage you to do the same..

  1. marc fawzi says:

    “(In particularly safe cities such as Fort Collins, Colo., that number can rise to 14 years — which is still no match for Google’s 300,000 miles.)” This is an improper comparison, statistically speaking. You’re comparing data from a very huge sample (all drivers in the US who have ever driven over the last 50 years, with total miles that are probably in the trillions, combined) to data from Google’s much smaller sample (1 computer algorithm over the last year or so covering 300,000 miles) You can make the comparison only when Google’s algorithm has been tested for an equal number of miles (trillions) and in conditions including ice on roads and heavy rain, and in all cities in the US. I don’t doubt that machines are more reliable than humans at deterministic logic (including driving) as well as probabilistic logic, but the Google algorithm has yet to be tested to the same level as the human drivers have been tested, so it’s hard to make a reliable statistical comparison.

Leave a Comment

Please keep your comments friendly on the topic.

Contact us