Last year I learned “machine learning” on coursera and found that it is very useful to develop self-driving car. This course was created in 2011. Since then, there has been much progress in self-driving cars. Last week I found two articles on self-driving cars. One is self-driving cars by google and the other is an autonomous truck. Let us see what they are and consider the impacts of these cars when they are available to us.
1. Self-driving cars
This is one the most aggressive project of self-driving cars because the goal of the project is cars without driver intervention. According to Google website, it says”a few of the prototype vehicles we’ve created will leave the test track and hit the familiar roads of Mountain View, Calif., with our safety divers aboard.”. It looks so small and cute. However, with computers and sensors, it can run without intervention by humans. I imagine machine learning is used to control self-driving cars as I learned it on coursera before. Because the machine can “learn” new things from data, the more self-driving cars run, the safer and more sophisticated they become. Therefore collecting many data on self-driving cars is critically important. I wonder when they can drive without drivers in future.
The other is autonomous trucks. According to Bloomberg, “Regulatory and technological obstacles may hold back the driverless car for decades. But one of the first driverless semi-trucks is already driving, legally, on the highways of Nevada.” This is a truck which can be controlled on highways. But in difficult tasks such as driving in parking lots, human should take over and drive them. It looks like “a truck, which is supported by computers”. Unlike self-driving cars by google, this truck needs human drivers. But it must be helpful for truck drivers when they drive on highways for long time.
3. What is needed to promote self-driving cars?
Firstly, we need to consider regulations about how self-driving cars are allowed to run in public. Because the more data is available, the more sophisticated self-driving cars become. In order to accelerate development of self-driving cars, data is like “fuel” to develop computers in order to control cars. Therefore regulations are very important to allow self-driving cars to run in the real world in order to collect data.
4. What are the impacts to our society?
In aging societies such as Japan, older people sometimes feel difficulties to drive a car to go to hospitals or shopping malls. In such a case, the self-driving car is one of the solutions for the problem. With self-driving cars, senior personnel can go anywhere they want without driving. In the emerging countries like Asean, a lot of trucks are needed to prepare the infrastructures and lifelines all over the countries. So it is very useful when self-driving trucks are permitted to run across country borders. Therefore, regulations should be considered as a region rather than country by country.
In the long run, we should prepare the shift from current situations to a digital economy. It means that some of jobs might be replaced by computers with machine learning. The more self-driving cars are available, the less truck drivers and taxi drivers are needed. Andrew Ng, the famous researcher of machine learning, talked about this shift on the article. “A midrange challenge might be truck-driving. Truck drivers do very similar things day after day, so computers are trying to do that too.”
No one knows exactly when self-driving cars are available in public. It does not look long-term future as I look at the development of technologies. We may have a lesson of self-driving cars. Andrew Ng says in the article, “Computers enhanced by machine learning are eliminating jobs long done by humans. The trend is only accelerating.”
What do you think?