Google's self-driving cars are hitting the city streets, handling the chaos well

By Andrew Martonik on 28 Apr 2014 03:01 pm

With a post on its official blog today, Google is giving an update on its self-driving car initiative, showing that the latest goal is to improve driving on city streets. Google's self-driving car project has already driven thousands of miles without a hitch, but until recently has primarily focused on simple highway driving. As anyone who has driven through a busy city core before knows, there are new and drastically more difficult to handle situations in the city than on an open road.

Pedestrians, bicyclists, traffic revisions, railroad tracks and downright bad drivers all throw additional variables at a computer that's trying to keep a car and passengers safe, but Google says that it's making big strides in managing these new challenges. Its latest self-driving car software can recognize dozens of common (and less-than-common) driving situations and immediately choose the best course of action. Whether it's a group of people crossing the street holding up a right turn or a disabled semi truck on the side of the road, the car now just knows what to do.

Google's self-driving cars becoming ubiquitous is still a good ways off, but we're starting to see a real refinement of what this technology is capable of doing. Just a little more time and testing is all that's needed now.

Source: Google

Andrew Martonik Andrew Martonik "West Coast Editor of Android Central" 33 (articles) 0 (forum posts)

Reader comments

Google's self-driving cars are hitting the city streets, handling the chaos well


I drive a 2012 jeep Grand Cherokee, it had advanced cruise control. It is radar in the front of the car that determines how far the car in front of me is and matches speed to a maxium speed I set and keeps an even distance. The Jeep will break and speed up as needed... it's wonderful for commuting. an hour on hwy and I don't touch the gas or break once. I have often thought, wonder how long until I don't have to steer, guess it's not that far away.

This is definitely cool progress to see. The main challenge, as Andrew pointed out, will be the complex and last-minute problem solving required on dense city streets.

Pedestrians who dash out after the light has already changed, anticipating what a bike courier will do versus someone commuting to work, or what to do when a street car comes to a stop (90% of the time it is to let passengers off an on, the other 10% of the time it is out of service/delayed so a driver can go around).

Not a problem that is impossible to solve, but definitely one that requires building a database of how the car should behave based on its location.

Love the way this has progressed. I wonder how the car would reach to a downed power line or a stray cow in the road... It just happened yesterday due to the storms in my area. I highly doubt they have those in the software already.