Who's driving this bus? Nobody

Commentary: Autonomous cars are all well and good. But if you want to make everyone's life better, look to your city's buses.

Commuting into a city can be a nightmare, am I right?

Maybe you're drumming your fingers through a traffic jam or squeezing onto an overcrowded train. Me, I live in London and I know how bad the underground system can get at rush hour. If the tube is delayed or being repaired, getting to and from the office can be the worst part of my working day.

But the next generation of public transport seeks to change all that -- by taking more individual cars off the road, self-driving shuttles and buses promise cleaner air, reduced noise pollution and safer areas to walk and ride.

Like many cities, London would grind to a halt without public transport. The city's iconic red buses support over 6.5 million journeys every day, and the underground subway network handles up to 4.8 million daily commuters. As London grows by a rate of 9-10 people per hour, it's essential to keep expanding and modernising the transport system to meet the city's increasing needs.

In recent years the transport authority has shortened queues by introducing contactless card payments that let passengers through turnstiles with less hassle. It has also begun to tackle fuel efficiency and air quality by putting fully electric buses on the streets and establishing new zones where only low-emission buses can operate. But a plan to bring driverless shuttles to the streets of London could be the most ambitious yet.

In April, the UK's Transport Research Laboratory (TRL) completed its first test of an autonomous shuttle in Greenwich, a part of Greater London. A prototype driverless shuttle carried members of the public through a 2km route around the Greenwich Peninsula, using built-in sensors and autonomy technology to avoid obstacles. It isn't all that different to a self-driving car, but it's far more beneficial to a city's ecosystem by ferrying more people than a single car.

"There's a philosophical difference between Google's two-passenger cars and putting lots of people on a single bus," Professor Nick Reed told me when I visited TRL. He's a technical lead on the Gateway project, working on the driverless shuttle trials.

• AUTOPLAY: ONOFF
• 0:00 / 0:15

"There's nothing wrong with a model that uses small vehicles, and they may be appropriate for some use cases," he said. But he argues that mobility in cities will only improve if we use larger vehicles that can move more people around at once.

The Gateway project shuttles are being tested in a designated lane alongside pedestrians and cyclists, not integrated with cars on the road. But once driverless vehicles are ferrying passengers alongside cars, we're likely to see changes to the way traffic flows through our streets.

You can make traffic more efficient if you can predict what it's going to do, Prof. Reed told me. And as cars become increasingly automated, they become more predictable. But he added that this doesn't mean we'll have to wait until every car is automated to see results. "The benefits [of driverless vehicles] appear quite quickly because the behaviour of other drivers is affected and constrained."

In other words, if the car in front of you is sticking to the speed limit, you'll have to do so as well.

The key? Don't drive like a human

Even if you're convinced that a driverless shuttle can handle itself on the road, how can it account for the behaviour of human drivers in other cars? The simple answer is that driverless cars will have to be more cautious than humans.

Click for more storiesAaron Robinson/CNET

"An automated vehicle is less likely to be in an unavoidable crash because it'll anticipate that happening and slow down before the situation arises -- which human drivers don't always do," Prof. Reed said.

If an accident isn't avoidable -- whether it's a collision, a near miss or something less serious -- the data from the incident will be analysed and can be used to rapidly update the automated system controlling the shuttles.

How can a computer anticipate a dangerous situation? Since 94 percent of collisions are caused by human error, it needs to understand how human drivers behave.

The TRL has been using a realistic car simulator since the 1960s to understand how we drive, and that simulator is now being used to test the way human and robot drivers will coexist on the roads. Test subjects "drive" through virtual Greenwich streets, and their reactions to real-life road encounters with driverless vehicles are monitored by researchers.

TRL has also installed driverless technology in human-driven cars as part of a separate project. By comparing the decisions an automated system would have made with the ones real drivers make in the same situations, we can better understand the differences between the way humans and machines conduct themselves on the road. This information will help driverless cars anticipate dangerous human driving and make those situations safer for their passengers.

We don't know yet how driverless shuttles will take to urban streets, but TRL's testing has already spotted a potential risk. While testing drivers' reactions to a simulated vehicle "platoon" on a motorway -- basically a herd of cars all moving together -- researchers found that humans imitated the behaviour of the platoon by driving too close to the car in front of them. The lesson: If we're going to let driverless cars set the pace, we have to make sure they're not a bad influence.

Cities are getting smarter

These plans aren't unique to London. A similar shuttle has already been launched in Las Vegas this year. Finland is also taking self-driving transport seriously, and began testing two autonomous buses last August. And Columbus, Ohio, is one of many cities considering the ways self-driving transportation can make life easier.

One of the self-driving buses tested in Finland last year.Vantaa-kanava fi; YouTube screenshot by CNET

Columbus was awarded $40 million by the US Department of Transportation for its winning pitch in the Smart City Challenge in 2016. Its proposed improvements include ride-sharing programmes and a single app that pulls together real-time data for pedestrians, drivers and people travelling by bus or cab. It will also deploy three self-driving shuttles.

In April, the mayor of Portland, Oregon, invited companies developing autonomous vehicles into the city, with the aim of using shared fleets of vehicles to reduce congestion and pollution.

Transportation Bureau Director Leah Treat told The Oregonian, "if we simply replace all of the cars on the road with driverless cars, we're not going to be any better off today."

Will bus drivers be out of a job?

As automation capabilities grow, fears about losing jobs to self-driving vehicles are understandable. The UK has already seen long-term strikes on the Southern Rail network when automation phased out train guards. These concerns have been echoed across other industries, and noted physicist Stephen Hawking has warned that artificial intelligence could lead to "entire industries disappearing."

Tesla and SpaceX founder Elon Musk speculated on the future of buses and the people who drive them. In his latest "master plan," Musk predicted that, "it will probably make sense to shrink the size of buses and transition the role of bus driver to that of fleet manager."

Automated shuttles and self-driving cars could coexist in cities.James Martin/CNET

Prof. Reed reminds us that we've already seen what happens when technology replaces humans in industries such as agriculture and manufacturing but points out that "productivity and employment in other roles has increased dramatically."

Will automated transport replace the human-powered system we have now? He predicts that "much will be complementary to existing modes of transport but, inevitably, some will be competitive." Automated transportation won't be introduced overnight -- it's likely to be a gradual transition, occurring at different paces and in varied ways around the globe.

In the meantime, the autonomous shuttles being developed in London and Columbus aren't intended to replace existing transportation and won't spell the end of traditional public transport. Instead, they're designed to tackle the inefficient "last mile" between, for example, getting off a train and reaching your destination. Instead of walking on a rainy day or clambering on a bus, you could hop onto a self-driving shuttle.

As for the passengers, they're less hesitant about automated transport than you might expect. "Once people find that vehicles behave in a way that they find acceptable, they begin to trust the automated system very quickly," Prof. Reed told me.

TRL has collected feedback about the project from people who live and work in Greenwich, which has so far been mostly optimistic. It seems that if automated technology really is safe and ready to be tested, people are happy to hop on board. Are you?