Google Creates Autonomous Cars
KITT
A Toyota Prius was going on the Highway 1 between San Francisco and Los Angeles a couple of days ago, but it was different than the other ones like it which might have passed on the same road. First of all, the car had a funnel-like cylinder on the roof. Secondly, no one was driving it.
Yes, you read it right, the car was driving by itself. There was a person inside of it, but he did not drive it. The car belongs to Google, and it is just one of the Google cars that belong to the Google Street View project. The main reason why it was able to drive on its own was thanks to an artificial-intelligence software program. The program is able to detect anything that is around the car, and to take certain decisions just like a human would. Seven such tests cars have been deployed on the roads from the United States of America. They have covered more than 1,000 miles without any human intervention and more than 140,000 with only occasional human control.
You should not think that the cars are left on the streets unsupervised. A technician sits in the passenger seat, constantly checking the software program, and a person is in the driver’s seat. If anything goes wrong and the error can not be controlled with the software, the person from the driver’s seat will immediately take control of the car. There was an occasion when the car drove by itself. The car managed to go on the Lombard Street in San Francisco, which is one of the curviest streets in the world. One of the cars got into an accident, but it was when it was hit from behind by another car when it was stationing at a traffic light.
Such cars will not be mass produced anytime soon. However, the technologists believe that they will be successful one day, and that they will become as popular as the internet currently is. The engineers have stated that the robot drivers are in a manner better than the humans. They have greater reflexes than the humans, they can “see” at 360 degrees, and they do not lose focus. On top of that, they do not make the mistake of exceeding the speed limit, or of driving while under the influence. They can be very useful, as they can save lots of lives. In the United States of America more than 37,000 people have lost their lives in car accidents in the year 2008 alone. The scientists believe that the technology could create roads which would have a double size, which would make it safer for the ones driving the cars because they would have more space for maneuvering.
Since the robot drivers would not crash, the cars would be much lighter, and as a result they would use less fuel than they use now. However, before any attempts of introducing the cars with robot drivers are made, they need to be 100 percent reliable. That will take decades. If you think about it, the technology behind the personal computers has been available for a very long time now, but they still crash once in a while. That is not too troublesome in the case of the computers, but in the case of the cars, that could mean death. The computers often get infected by viruses, and the car computers could suffer the same fate as well. The fact that Google is investing in such technology is proof that the company is interested in more than remaining the number one search engine in the world.
Last week, a Toyota Prius equipped with all sorts of gadgets entered the Highway 101, which is the freeway through Silicon Valley. I should say that the Highway is very fast, which meant that in case of an accident; the consequences would have been great. The car kept the speed limit, as it has the speed limits for all the roads from the United States of America in its database. It followed a pre-established GPS route, and the device which was located on top of it, constantly updated the map. The car then drove on the highway, paying close attention to all the stopping lights, traffic signs, and so on. According to the ones who created the program, the software has different “personalities”. It could be aggressive which means that it would not stop and let someone pass in front of it, or it could be calmer, situation in which it would let the other cars pass it in certain situations.
Even in the aggressive mode, it would not put the ones inside the car in danger. Christopher Urmson, a Carnegie Mellon University robotics scientist, was the one in charge with taking control of the car in case something would go wrong. In order to do that, he had to do two things: push a red button, then move the wheel, or hit the brakes. He intervened twice: first when a bicyclist ran the red light, and the second time when a car suddenly stopped and backed into a parking spot. It seemed that the car had control of the situation though, as it began to slow down when the incident with the bicyclist occurred. After that, the car was put back on to automatic mode, and the two persons from the car were able to communicate freely.
Urmson stated that the car drew lots of attention, as the people were amazed by the fact that no one is driving it. The program was created by Sebastian Thrun, the 43-year-old director of the Stanford Artificial Intelligence Laboratory, who also invented the Street View concept. He won the second Grand Challenge of the Defense Advanced Research Projects Agency for driving autonomously over 132 miles in the desert with the Stanley robot car. He was helped by the Stanford students with the design of the car, and they received $2 million from the Pentagon for taking the second place. More than 12 people were hired by Google, and they have the task of sitting in the driver’s seat in order to intervene in case something goes wrong. These persons are paid $15 per hour and sometimes even more.
The Google researchers stated that it is unknown what the fate of the current program will be. They have no idea if such cars will be used for the Street View program or if they will remain to the regular cars. Dr. Thrun said that Google is willing to invest in technology which might be usable only in the future years. It is almost 100 percent that this technology will not be used in the next eight years or so, and still Google is willing to invest tens of millions into the technology. Google might use the technology to provide information to the makers of autonomous vehicles, or they could sell the technology to the higher bidder. The situation is complicated for the moment, because under the U.S laws, a person must be in the control of the car at any moment. The reality is that people might make mistakes, and they often do in most of the cases. Since a person can take control of the car at any moment, the Google cars are 100 percent legal. Such cars have been designed since 1960, but the breakthrough came in 2004.
If a human is controlling the car, he might not pay attention to a school zone, and he might hit a child. The software program would have paid attention to the area, and the accident would have not happened. Would such a situation create any changes, or would the laws and regulations remain the same? In case of an accident, who would be responsible, the man behind the wheel, or the one who designed the software program? Bernard Lu, senior staff counsel for the California Department of Motor Vehicles, stated that there are many faults with the laws regarding the use of a vehicle in the United States of America. A contest was held in 2005, and it involved autonomous cars.
Dr. Thrun’s Stanford team built the car that won the race, but the competition was harsh, which proved that the technology behind such vehicles evolved a lot. Dr. Thrun believes that such cars are superior to any other car which exists nowadays. They would be able to consume much less gas than a regular car, and they would reduce the number of accidents which take place on the roads. He believes that in 100 years or so, the situation could be completely different. For example over that period of time, common cars might exist, which might be used by numerous people. One would have to “ask” for it, and the car would come to his destination. Of course, this sounds like science fiction, but the reality is that autonomous cars similar to the ones used like Google in their projects, seemed science fiction a few years ago as well.
