By now most people have heard about Google's secretive Google X Lab, where untold tales of amazement are being brought to life. One of those initiatives, is the "driverless car" – a car that could literally drive itself, with no interaction from its human driver…and eventually no driver. Say you came out of work only to find you missed your last bus and you have no money for cab fare – just use your smartphone to summon your car to come and pick you up! Sounds like pretty wild stuff, like something from a sci-fi movie, but the reality of it is – it could soon be reality!
Google has been working the past few years on their self-driving car project, as have a couple of car companies themselves, and Google's driverless cars have been taking rides on the highways for thousands of miles. However, the real trouble starts when you want this driverless car to maneuver in city situations where there are so many 'distractions,' they would need to avoid, including jaywalking pedestrians, cars lurching out from hidden spots, doubled parked vehicles, traffic lights, 4-way stop signs, road construction, and on and on.
This is why Chris Urmson, Director, Self-Driving Car Project, posted on Google's Blog that the team was directing their attention to mastering city driving. As he explains that a mile of city driving is much more complex than a mile of highway driving. There can be hundreds of different objects moving to a set of their own rules. They have improved their software to be able to recognize and detect the many possibilities that could arise during a drive in the city – to the point where it is so much more in tune to the car's surroundings than a human brain could ever comprehend – monitoring hundreds of distinct objects simultaneously. The self-driving car with its computer never tires or becomes distracted as a human driver would naturally become. Urmson adds, "As it turns out, what looks chaotic and random on a city street to the human eye is actually fairly predictable to a computer…Our vehicles have now logged nearly 700,000 autonomous miles, and with every passing mile we're growing more optimistic that we're heading toward an achievable goalâ€"a vehicle that operates fully without human intervention."
The radar and laser sensors can create a 3D map of the car's surroundings in real-time, and then Google's software sorts them into four main categories – moving vehicles, pedestrians, cyclists, and static things, such as different signs, parked cars, or curbs. In the past, the software would detect a group of individuals as one person, but now it is so refined that it can distinguish the individual people. Google originally thought that it would have to program in all of the world's stop signs, but now the software onboard can recognize a stop sign to the point of even recognizing a sign held up by a crossing guard.
There are other problems to work out – for instance, because the cars are designed to drive defensively, the car knows when to stop at a 4-way stop sign, but hesitates to pull-out on its turn, as it waits for the other cars to get out-of-the-way. Even allowing the car to turn on a red light after it stops is another problem…but certainly, these are small problems considering what they have accomplished so far. Google seems confident that we will see these vehicles in everyday use by 2017.
Check out the interesting video below – you will gain a little insight to how Google is attacking the problems of a self-driving car. Please join in a discussion on our Google+ Page about these self-driving cars and what you think about them…as always, we appreciate your opinions.