Here's how Google's driverless car avoid pedestrians, bikers on the streets (at least most of them)


A driverless car in the city streets -- previously it sounded fictive, much like vehicle in a sci-fi movie, but Google is gradually proving it is real and possible in a highly technical driven world.

Google's driverless car project dates back in 2009, when the early days of the driverless car were mostly in the confinement of the highways in California. Now, the driverless car is out and about with 700,000 recorded miles up on its sleeves, mastering the twists and turns in the fast lane.

"That's why over the last year we've shifted the focus of the Google self-driving car project onto mastering city street driving," Chris Urmson, director of the Self-Driving Car Project, wrote in a blog post.

He joined the company in 2009 to develop the driverless car because he thought it might "change the world."

"You make so much more progress when you're thinking about changing the world rather than making this minor delta improvement on something," he said, adding that the thought can get one all fired up in the morning.

Urmson shares that the self-driving car appears to be managing how to stay in its lane and keep its speed, eventually encouraging the team to take the car out of the freeway and test in the erratic behavior of city streets. He also reveals they enhanced their software for better detection of multiple diverse objects simultaneously such as buses, pedestrians, stop signage and gestures of cyclist among others.

Based on the video shown in the blog post, the driverless car appears to have the sense of a human to pay attention and react to all these factors. It understands the gestures of the cyclist who was about to make a turn. It also recognizes the obstacles ahead of the road such as construction zones, changes lane when it senses road blockages and distances itself from large objects such as trucks.

"As it turns out, what looks chaotic and random on a city street to the human eye is actually fairly predictable to a computer," said Urmson.

The only difference is, of course, the self-driving car never gets distracted or tired like humans.

There were few instances when it seemed to be slow in responding or understanding the scenario around, but the developments are still impressive. Urmson admits they still have to figure out many issues such as teaching the vehicle to drive other streets in Mountain View, but what used to be roadblocks for the vehicle can be navigated freely now.

He says that his biggest goal on Google's self-driving car project is safety and then cites statistics on deaths caused by road accidents, including human error. He says there are 33,000 deaths yearly on U.S. roads, with car crashes as the leading cause of the deaths of people and that 90 percent of the collisions were brought about by human error, which makes this project a "big deal" for him. He and his team think they can make one that is really much safer than human driving.

To view how the self-driving car of Google performs in a much busy street, check out the video below.


ⓒ 2018 All rights reserved. Do not reproduce without permission.
Real Time Analytics