Sunday, October 23, 2011

How to Self-Driving Google Car Works

Once again a secret project, Google's autonomous vehicles (autonomous vehicle Google) has come out, openly with the company's test-driving them on public roads and on one occasion, even inviting people to ride in one car robot that shot at close range just .

Google's fleet of Toyota Priuses Robotic has now recorded more than 190,000 miles (300,000 kilometers), driving in city traffic, a busy highway, and the mountain road with only occasional human intervention. The project is still far from being commercial, but Google has set up a demonstration system on campus, using a golf cart without a driver, which points to the way how technology can turn into transportation, even in the near future.

Stanford University Professor "Sebastian Thrun", which guide the project and engineers Google "Chris Urmson" that discusses this and other details in a keynote address at the IEEE International Conference on Intelligent Robots and Systems in San Francisco last month.

Thrun and Urmson explains the workings of the car and showed a video of the test drive, including footage of what is "seen" on a computer on-board and how to detect other vehicles, pedestrians, and traffic lights.



Urmson, who led the technology for the project, said that "the heart of our system" is a laser range finder mounted on the roof of the car. Apparatuses, a 64-VELODYNE laser beam, producing detailed 3D maps on the environment. The car was then combines laser measurement with high resolution map of the world, producing various types of data models that allow for the drive itself while avoiding obstacles and respecting traffic laws.

Vehicles also carry other sensors, which include: four radar mounted on the front and rear bumpers that allow the car to "see" far enough to be able to deal with fast traffic on the highway, the camera is positioned near the rear-view mirrors that detect traffic lights , and GPS, inertial measurement units and an encoder wheel that determines the location of the vehicle and track its movement.

Two things tampakyang very interesting about Google's approach. First, it relies on very detailed maps of roads and terrain, Urmson says that it is very important to determine accurately in which the car. Using only GPS-based techniques.

The second thing, before sending a self-driving car on the road driving test, engineers Google first drove along the route of one or more than once to collect data about the environment. When the autonomous vehicle's turn to drive himself, he will compare the data obtained to the data previously recorded, an approach that is useful to distinguish pedestrians from stationary objects such as poles and mailboxes.

Video above shows the results at one point, where you can see a car stopped at an intersection. After the light turns green, the car started to turn left, but if there is a pedestrian crossing would not be a problem because the car will stop for pedestrians, and even for those who decide to cross at the last minute.

Sometimes and somehow, the car should be more "aggressive." When going through a four-way junction, for example, overtaking other vehicles based on the rules of the road, but if the other car did not return, it is little progress to show his intentions to other drivers. Without such programming such behavior, it is impossible for cars to drive the robot in the real world, Urmson said.

The Google engineers have a lot of fun when viewing fast forward to the speed of cars through the Google parking lot, tires squealing at every turn with a time of 13 minutes, and Urmsonpun smiling broadly.

But the project has a serious side. Thrun and his Google colleagues, including co-founder Larry Page and Sergey Brin, believe that this smart vehicle could help make transportation safer and more efficient. This car will be pushed closer to each other, making better use of 80 percent to 90 percent free space on the road, and also set up a fast convoy on the highway. They will react more quickly than humans to avoid an accident, potentially saving thousands of lives. Making intelligent vehicles will require a lot of computing power and data, and that is why it makes sense for Google to go back to the project, said Thrun.

Urmson describes another scenario they imagined. Vehicles will be a shared resource, a service that people will use when needed. You've just touched on a smartphone, and an autonomous car will appear where you are, ready to drive you anywhere. And you will just sit and relax or do work.

He said that they collect video that showed a Caddy concept called Beta that shows the idea of ​​shared vehicle - in this case, an autonomous fleet of golf carts. He said the golf car is simpler than Priuses in terms of sensors and on-board computer. In fact, the train can communicate with sensors in the environment to determine their location and "see" incoming traffic.

"This is one way we look at future technologies. It really makes better transport, making it more efficient," Urmson said.

0 comments:

Post a Comment

Lowongan Kerja Terupdate

 
Design by Betr4y3r5 | Bloggerized by Free Blogger Templates | http://cafeinfo-net.blogspot.com/