Experts expect self-driving cars to make the roads much safer, and driving much more convenient. But what will they do to the environment? A newly published study suggests that, under some scenarios, the shift to autonomous vehicles could double energy use and greenhouse-gas emissions.
The good news is that other scenarios could lead to a nearly 50 percent reduction in those metrics by 2050, which would brighten the picture for coping with climate change. It all depends on how driverless cars are introduced into the marketplace, and how consumers and businesses respond.
“There is lots of hype around self-driving cars, much of it somewhat utopian in nature. But there are likely to be positives and negatives,” University of Washington engineering professor Don MacKenzie said. “By taking a clear-eyed view, we can design and implement policies to maximize the benefits and minimize the downsides of automated vehicles.”
MacKenzie is one of the authors of a study analyzing the range of possibilities, published today in the journal Transportation Research Part A. The survey comes as a plethora of companies, ranging from Ford and Tesla to Google and Apple, are hustling to make vehicles more autonomous and jump through regulatory hoops.
The rise of driverless cars
You can already buy cars with some autonomous features, such as active lane-keeping, autopilot and auto-parking. Google is now testing its self-driving prototype on the streets of Kirkland, Wash. Peterbilt Trucks, a subsidiary of Bellevue-based Paccar, has been testing a “platooning” system that electronically links up trucks in a highway convoy for the sake of fuel efficiency.
Experts predict that fully autonomous vehicles will make their appearance on roads over the next five to 20 years – at first under limited conditions, and then more widely. David Jumpa, chief revenue officer for Seattle-based Airbiquity, told GeekWire that driverless vehicles are likely to make an early impact on freight deliveries. Why? Because the trucks follow routes on the transportation grid that lend themselves more easily to automation.
Rideshare services such as Uber and Lyft are also likely to face disruption, just as those services disrupted the traditional taxi industry. GM and Lyft are already talking about testing a robo-taxi system in Austin, Texas.
“If I wanted to crash that business model, what’s the most expensive part of the service? The driver,” said Jumpa, whose company develops technologies for connected cars.
Crashing existing business models could crash employment patterns as well. This month, Rice University computer scientist Moshe Vardi said rapid automation, particularly in the transportation industry, could leave half the world’s population without a job by 2050. He argues that the trend has already begun in Australia, where two iron mines have gone completely over to autonomous trucks for ore transport.
“We will be almost fully automated in 25 years,” Vardi said.
Impact on the environment
The way Vardi sees it, the switch from human-operated machines to autonomous machines will be such a game-changer that it should be a top item on the public policy agenda. That goes for environmental policies as well as economic policies, MacKenzie told GeekWire.
Light-duty vehicles, and the processes required to manufacture and fuel them, are thought to account for a quarter of U.S. greenhouse-gas emissions, he said. Significant changes in costs and consumption patterns could have a big impact on climate change, for good or ill.
“We need to remember that automation is a means, not an end. … Automation is not going to save us, and it could make things worse,” MacKenzie said. He and two colleagues, the University of Leeds’ Zia Wadud and Oak Ridge National Laboratory’s Paul Leiby, quantified the potential effects of automotive autonomy over the next 35 years to get a sense of what’s at stake.
Their analysis included some factors that have gotten less attention in the debates over driverless cars. For example, self-driving cars might put more elderly people on the roads, because they won’t be hampered so much by the disabilities that come with age. Vehicles might eventually be made of lighter, more fuel-efficient materials, because crashworthiness would become less of a concern. Travelers might be more prone to use their cars and less willing to use public transit, trains or planes, because they could be working or relaxing in comfort while they ride.
The study lays out a range of scenarios rather than specific predictions. If all the fuel efficiency benefits kick in, with a minimal increase in total miles traveled, energy demand could decrease by as much as 45 percent. But if the benefits don’t materialize, and if travel demand and highway speeds increase, the energy demand could more than double.
“Such a scenario is highly unlikely, given other constraints that will possibly limit such an increase, but it is still useful to highlight the potential increase in energy demand due to automation,” the researchers write.
Which policies will work?
MacKenzie and his colleagues advise governments and businesses to boost policies that accentuate the positive, and minimize if not eliminate the negative.
The researchers say many of the positive effects can be delivered by systems that still require the human driver to pay attention to the road. They contend that those systems are less likely to bring about the negative environment effects associated with fully autonomous systems, such as higher levels of vehicle usage.
Unfortunately, other experts say semi-autonomous vehicles wouldn’t be as safe. Google has repeatedly argued against dividing the driving responsibilities between the human and the machine. “This middle ground could be a little bit dangerous,” Demis Hassabis, a co-founder of Google DeepMind, said at a conference this month.
As we all negotiate the middle ground, MacKenzie provides a few recommendations for accentuating the positive, plus one recommendation for minimizing the negative if necessary:
Promoting platooning: A system that allows vehicles to travel down the highway in a convoy, also known as platooning, offers clear fuel-saving advantages by reducing air resistance and minimizing traffic congestion. But the vehicles would have to be linked together electronically in order to derive the maximum benefit. MacKenzie said “it may be necessary for the government to standardize communication between vehicles,” so that semi-autonomous vehicles of different types can join in the same convoy. The U.S. Department of Transportation is currently working with academic and industry partners, including Peterbilt, to study the options for platooning.
Encouraging eco-driving: Eco-driving techniques (for example, maintaining a steady speed at low engine RPM) can save up to 15 percent in fuel costs. Most drivers aren’t in the habit of applying those techniques, however. “Consumers notoriously undervalue new technology for fuel efficiency,” MacKenzie said. Eco-driving algorithms could be built into the software for autonomous vehicles, and it’d be even better if regulators provided more incentives for carmakers to do so, MacKenzie said.
‘Right-sizing’ robo-rides: If ride-summoning services make the switch to self-driving cars, that could increase the opportunities for “right-sized” travel, MacKenzie said. For example, Robo-Uber could send out a super-compact car to give Mom a quick ride to the office, or a self-driving minivan if the whole family wanted an outing. “People would no longer feel forced to buy a car for their worst-case peak requirements,” MacKenzie said. Robo-cars could also turn into mini-transit services, picking up multiple riders along a route a la Shuttle Express. But regulators would have to give the OK. If Uber is already running into resistance in places like Vancouver, B.C., just imagine how much trouble Robo-Uber could face.
Resorting to road pricing: If robo-rides catch on in a big way, MacKenzie and his colleagues say increased vehicle usage and congestion could more than cancel out the improvements in energy efficiency. As Airbiquity’s Jumpa noted, the driver’s time and trouble account for the biggest “cost” of automotive travel. Eliminating the human driver effectively reduces the travel cost, either financially or psychologically. “When you make something easier, cheaper and more convenient, you should expect people to use more of it,” MacKenzie said. The so-called “Law of Traffic Congestion” claims that traffic expands to fill added capacity, and there’s actually research to support that view.
So what can be done to restrain the rise of robo-traffic jams? “It’s never popular, but if we’re talking about reducing the biggest cost of driving, we probably should be thinking about road pricing to replace that reduction in cost,” MacKenzie said. The software and communication capabilities in a self-driving vehicle would make it relatively easy to collect a per-mile highway toll.
Bottom line? If you dislike paying the automatically collected tolls on Interstate 405 or the floating bridge on State Route 520, you might well come to hate the rise of the self-driving, tax-collecting machines.
Check out this policy paper on self-driving cars from the University of Leeds’ Center for Integrated Energy Research.