2 . Waymo vehicles are testing so frequently in our neighborhood that we usually see four of them on our 15-minute dog walk. My daughter Carson likes Waymo because they look distinct. The fact that a robot is driving, however, fails to impress her. This is just the reality she has been born into.

Already Waymo vehicles are in motion for San Francisco to be the first major city with a driverless ride-hailing (打车服务) service.

Yet I remain doubtful, It’s not so much the technology. When I arranged for Waymo to pick up my family for a 20-minute test ride, the robot made no obvious mistakes. And in the “truly driverless” ride-hailing service Waymo has been operating for more than two years in Phoenix, it has yet to be involved in any major incidents. No, the reason these services aren’t widely available is mostly to do with the absence of a clear business model.

Back in 2014, Travis Kalanick, then-CEO of Uber, said the prospect of driverless cars posed an existential challenge to Uber. Human drivers account for two-thirds of the cost of an Uber journey, he reasoned, so “when there is no other human in the car, the cost of taking an Uber anywhere is cheaper”.

Since then, it’s become clear that the economics are more complex. What made Uber so disruptive is the gig (零工) economy. The app serves nearly 100 million people each month in more than 10,000 cities, without Uber owning a single car. Its drivers are contractors, so it only pays them when they are carrying passengers. If the driver is waiting around at the airport, say, it’s not Uber’s problem. It has sophisticated software to match riders with drivers, and to do it quickly.

The driverless industry throws away these advantages. Instead of hiring drivers on the cheap, Waymo needs to build or purchase vehicles costing some $200, 000 apiece. Owning too many cars and they are wasting capital. Too few and they’ll be frustrating passengers with lengthy wait times. A driver also plays the role of conductor, ensuring the car is fueled and avoiding illegal activity. Driverless cars will have to be serviced and monitored to do the same thing. “Driverless does not mean humanless,” says Ashley Nunes, a researcher at MIT. His work on the subject has concluded that robotaxis can’t compete with the gig economy on price. “Robotaxis replace one set of human costs, the human driver, with another, inefficiency,” he adds.

When robotaxis finally open to the public in San Francisco, they risk becoming a gimmick (噱头). Fun to try, sure, but they’ll cost the same as an Uber, arrive no faster and decline to speed up when you’re in a rush.

3 . On average, cars sit, doing nothing, 96 percent of the time. That makes them ideal candidates for the sharing economy. The potential to reduce traffic jams is enormous. A handful of car-sharing systems are already having a major impact on the total number of vehicles in our cities. Scholars have estimated that every shared vehicle removes nine to 13 privately owned cars from the streets.

The benefits will grow greatly as autonomous vehicles, that is, self-driving cars, currently available in experimental forms, gain a notable portion of the market. “Your” car could give you a lift to work in the morning and then, rather than sitting in a parking lot, give a lift to someone else in your family—or to anyone else in your neighborhood or social media community.

As a result, a single vehicle could go from one to 24 hours of use a day. A recent paper by our colleagues at the Massachusetts Institute of Technology report s that, under such conditions, the mobility demand of a city like Singapore could be met with only 30 percent of its existing vehicles. In addition to vehicle sharing, autonomy could open up a new wave of ride sharing. Already applications such as Via, uberPOOL and Lyft Line allow different people to share the same ride, cutting operating costs and individual fares. Autonomy could boost ride sharing even more because all trips could be managed online. In cities, the potential for ride sharing is significant, based on analyses by our Sense able City Lab at M. IT.

New York City, for example, is obviously shareable. Our lab’s HubCab project gathered data from 170 million taxi trips involving 13,500 taxis in the city—specifically, the GPS coordinates (协调) for all pickup and drop-off points and corresponding times between the two. We then developed a mathematical model to determine the potential effect of ride sharing applied to those journeys. The project introduced the concept quantitative results revealed how taxi sharing could reduce the number of cars by 40 percent with only rainimal delays for passengers.

Combine car sharing and ride sharing, and a city might get by with just 20 percent the number of cars now in use, with its residents traveling on-demand. Of course, such reductions are theoretical. In real life, they would depend on how willing people are to share rides and adopt self-driving technology. But any drop in the number of vehicles could lower the costs and energy associated with building and maintaining our mobility infrastructure (基础建设). Fewer cars might also mean shorter travel times, fewer traffic jams and a smaller environmental impact.

Baby Steps Toward Autonomy…

Companies, like Google and Tesla, didn't even exist when Toyota introduced the Prius in1997. However, they have become major     1     in the auto industry recently. Both of them aim to introduce fully autonomous cars ---that is, cars that drive themselves---within the next several years.

Self-driving cars are expected to be much safer than human-driven ones. But even if the first robot cars     2     the roads in the next few years, most of us probably won't give up driving entirely for at least another 15 or 20 years. In the meantime,     3     cars will gradually take over certain aspects of driving.

Companies have been adding     4     features to cars since the 1990s---things like adaptive cruise control, which uses sensors to adjust a car's speed based on the traffic in front of it, and automated parallel parking. Some cars automatically stop—or at least slow down—if a driver doesn't step on the brake in time to avoid a crash. In a certain model developed in 2017, the driver will be able to change     5     simply by hitting the turn signal for two seconds (the car will take care of the rest). Within a few years, cars may be able to determine when an accident is likely and make     6    to the cabin---moving seats, closing windows , retracting the steering wheel.

Even better than preparing for a crash, of course, is     7    one. Some vehicles     8     warnings when they detect, through cameras and sensors, that a driver is getting sleepy. Future cars might take over for     9     drivers---or automatically pull to the side of the road and shut down. Biometrics(生物识别技术）could aid this process. If a car has sensors that can measure a driver's breath and heart rate, it could     10     into self-driving mode when a driver has a heart attack.