Self-driving robotaxis are taking off in China, but how do they work?
Automated vehicles are a hot topic in the media because they are predicted to lead to a reduction in traffic injuries and deaths, and to a dramatic increase in productivity and speed. But there’s one big risk: the risk that such vehicles will be hacked, and used by a sophisticated criminal to avoid detection or for their own purposes. Last year, police in Japan were forced to release footage of a man driving past a police car that was fitted with an automated license plate reader, with the goal of identifying the owner. The man was later arrested.
This is a potential worry in China. More than 1 million automated cars have been sold to Chinese users; by 2020, a quarter of them could be self-driving. “You might as well call them robots,” says David Eger of MIT’s Center for Transportation and the Automobile. “They don’t really have wheels or anything else, and they can do pretty much anything you can do, with a little extra driving. In many cases, they look like regular taxis.”
Self-driving cars have yet to reach widespread public use. To do so, they will need to be deployed at a large enough scale to be able to detect pedestrians, cyclists, and other vehicles in need of traffic help. According to Eger, who also works on autonomous vehicle research at MIT, a company like Uber would probably try to do that by building a network of cameras around the city, but a car’s range of vision is not great. If no one was around to see it, it wouldn’t be able to see anything.
The most efficient way to build an array of cameras is by a system called radar. The car looks at the distance between the car and the radar, looking for that distance to decrease through the day to eventually zero—and then it automatically makes the turn. It is like a radar system