The following is an excerpt from the book, Driverless: Intelligent Cars and the Road Ahead by robotics researcher Dr. Hod Lipson and writer Melba Kurman, available from The MIT Press.
PCMag recently welcomed Dr. Lipson as a guest on our video/podcast interview series, The Convo (full episode embedded below). Dr. Lipson is the director of Columbia University's Creative Machines Lab, and he spoke at length about the new frontier of self-aware, self-replicating, and even creative machines. We talked about how these newly enabled robots in general—and driverless cars in particular—will reshape the economy and the world in the decade to come.
One of the great unknowns about driverless cars is whether their convenience will worsen traffic congestion and its accompanying evils. An optimistic scenario would be that driverless cars will improve the efficiency of urban transportation systems, and hence reduce private vehicle ownership, thereby reducing congestion and therefore reducing the size of a city's carbon footprint that's related to transportation. Another, less environmentally friendly scenario is that as people embrace the convenience of friction-free mobility, driverless cars will wind up actually logging more vehicle-miles per year on average, leaving a larger carbon footprint.
Convenience can be a double-edged sword. People are drawn to convenience like iron to a magnet. Sometimes, however, convenience carries with it a price: Unexpected and negative consequences. The friction-free personal mobility offered by driverless cars might solve the worst excesses already inflicted on us by automotive technology. Or, the hidden cost of convenient personal mobility might be that an ever-growing number of people casually rack up their number of miles driven.
Economists call the unforeseen reduction of expected gains from new technologies owing to increased usage the rebound effect. It's not clear whether driverless cars will have a rebound effect on traffic, increasing the number of miles that people travel each year, and the number of cars on the roads. Some research paints an optimistic picture, in which city streets will be emptier of vehicles in a few decades.
Yes, that's a giant "self-aware" robot spider in the center.
In an interview with The Economist, Luis Martinez of the International Transport Forum, a think tank dedicated to transportation policy, predicted that fleets of self-driving vehicles could replace all vehicular public transportation taxi and bus trips in a city, providing as much mobility but with far fewer vehicles.
To test this theory, Martinez created an agent-based model to simulate daily travel patterns in a medium-sized European city. Using several years of actual data from previous transportation surveys, he calculated that if city inhabitants used fleets of shared autonomous taxis rather than privately owned cars and public transportation, the number of vehicles on the city's roads could be reduced by 90 percent.
While fleets of autonomous taxis would drastically reduce the number of cars on the streets, the simulation also predicted that the overall number of vehicle-miles traveled per car would increase slightly because the self-driving taxis would shuttle back and forth more frequently to pick up passengers.
A report from the University of Michigan Transportation Research Institute supports these findings. The report concludes that the adoption of autonomous vehicles would reduce the number of cars owned by the average U.S. household from just over two to one vehicle per household. According to the report, one-vehicle households will be made possible because self-driving vehicles will use a "return-to-home" mode after they drop one household member at work so other household members can use the family self-driving car to be shuttled to errands and activities.
There's a catch, however. Although a family's driverless car can transport family members efficiently back and forth, the fact that one car is supporting more people would result in higher per-vehicle mileage. Although the average household of the future might own fewer cars, the remaining driverless car will be used 75 percent more frequently, accumulating an average of 20,406 annual miles per vehicle per year. The upside of this finding is that even if a single driverless vehicle were to rack up 75 percent more miles on average, the mileage for the entire household would still be lower than if two human-driven cars were in use.
One potential risk of having a single driverless car support an entire household is that the increase in per-vehicle mileage ends up being more than the predicted 75 percent. There's no doubt that summoning a driverless car to pick you up and drop you off would be a great convenience. However, an unintended negative consequence of more efficient transportation could be that a driverless vehicle will drive significantly more miles than would the equivalent human-driven vehicle.
Ideally, an empty self-guided car would find a safe place out of the way of traffic to sit and await its next summons. If that safe place were several miles away, however, the car would be forced to drive itself back and forth a great distance rather than just parking nearby. Its mileage would increase, and its wasteful shuttling would make traffic congestion and air pollution even worse.
If the availability of too-convenient transportation creates a rebound effect on traffic and dramatically increases the number of road miles that people travel each year, driverless cars could have a devastating environmental impact. Today the transportation sector is already one of the largest contributors to air pollution. In the United States alone, exhaust from cars and trucks causes an estimated 29 percent of the greenhouse gas emissions that human activities generate each year. If driverless cars were to increase the number of vehicle miles traveled per capita, densely populated "megacities" in developing nations would be hit particularly hard.
Google may have been the first to show a driverless car to the public, but they're far from the only one working on it.
While the United States has a nearly 100-year-old relationship with the car, other nations are enthusiastically catching up. China is following in the footsteps of the United States, gaining its own car culture. As a growing and newly affluent Chinese middle class embraces the convenience of car travel, cities such as Beijing and Zhengzhou are suffering from spectacular eight-lane traffic jams and worsening smog levels.
Today, the ratio of cars per person is still lower in China than in the United States or Europe, averaging 85 vehicles per 1,000 people (compared to 797 vehicles per 1,000 people in the United States). However, the rate at which the Chinese auto industry manufactures and sells new cars continues to skyrocket, increasing at an annual rate of 7 percent since 2013.
Perhaps Chinese car culture will sidestep some of the worst excesses of car culture by adopting driverless cars sooner, rather than later. To tame the traffic beast, Baidu, the Chinese search engine company some describe as the Google of China, is working together with BMW to develop autonomous vehicles that are familiar with Chinese roads.
- READ: When the Internet Takes the Wheel
In both developing and developed nations, traffic jams are a major source of air pollution. In the United States alone, as commuters inch forward in traffic jams, their idling cars waste 2.9 billion gallons of gasoline each year, enough to fill four football stadiums. Only time will reveal whether driverless cars will produce less pollution, or whether their use will entice people to log an ever-growing number of miles each year, further degrading air quality and making urban traffic jams even worse.
Another environmental side of effect of driverless cars could be shorter vehicular lifespans. A car's longevity is indicated by its odometer. According to Consumer Reports magazine, today the typical life span for a personal vehicle is about 150,000 miles which means that on average, over the course of eight years, that car will be driven about 18,750 miles per year. In comparison, since it drives roughly 70,000 miles a year, the lifespan of the average New York taxi cab is only 3.3 years.
It remains to be seen whether the introduction of driverless cars will ease the negative effects inflicted on us by the modern automobile. If the University of Michigan research is correct and a driverless car racks up 20,406 miles each year, the average family car would be "used up" more quickly, reaching its lifetime expectancy of 150,000 miles in just over seven years of use.
Uber, which has yet to make a decent profit is depending on self-driving taxis to justify their business model.
One worst-case scenario would be a future in which used-up driverless cars litter the landscape, filling junkyards and backyards with decommissioned auto bodies and worn-out engines. History has taught us, however, that new technologies do not merely extend a former status quo. Driverless cars have several characteristics that could change their potentially gloomy and environmentally devastating trajectory.
If the internet of the 1990s were suddenly forced to absorb today's data traffic, it would buckle under the load. Over the years, several enhancements have enabled the modern internet to absorb new users and handle an increasing amount of data, including better compression technologies, fiber-optic cable, and more intelligent routers. Similarly, improvements in technology could also ease the potentially negative rebound effect caused by driverless cars. Several research studies support such an optimistic view.
First, let's address the issue of vehicular lifespan. A report from McKinsey calculates that driverless cars will be able to brake and accelerate more gradually, resulting in fuel savings of 15 to 20 percent and a reduction of CO2 emissions of 20 million to 100 million tons per year. If McKinsey's research is correct, then smoother driving would increase a driverless vehicle's longevity.
Not only would driverless cars last longer, they could be built specifically to achieve longevity. There's nothing sacred about a lifespan of 150,000 miles. If there were a market for it, car companies could design driverless cars that could drive for several hundred thousand miles. City transit operators expect their buses to have a useful lifespan of at least twelve years and 250,000 miles. Semitrailers are designed to operate for 1,000,000 miles and their engines are designed to run virtually nonstop. Rail cars last even longer: some of the original BART cars in San Francisco, built in 1968, are still in operation today.
Even if their lifespan remained the same as today's human-driven cars, driverless cars could milk more capacity out of existing roads. To decrease their wind resistance, cyclists ride behind one another in a closely spaced line, an energy-saving strategy known as drafting. Fleets of driverless cars and trucks could use a similar approach and save energy by driving behind one another in tight formation, a fuel-saving strategy known as platooning.
Platooning saves fuel both by reducing wind resistance and using road "real estate" more efficiently. Human-driven cars don't use the space on the road very efficiently. People have to drive several hundred feet apart for safety and we aren't very adept at smoothly changing lanes. In contrast, platoons of driverless cars would use road space more effectively, resulting in less congestion at the places where traffic jams regularly form, such as highway on-ramps and off-ramps, before lane changes, and at intersections.
A study by researchers at the University of Texas estimates that if 90 percent of the cars on the road in the United States were self-driving, it would be equivalent to doubling road capacity. Texas researchers predict that tightly spaced platoons could reduce congestion-related delays by 60 percent on highways and by 15 percent on suburban roads. Trucks, because of wind resistance, are particularly prone to fuel inefficiency. Platoons of autonomous trucks spaced fewer than three feet apart while driving would reduce fuel consumption by 15 to 20 percent per truck.
One of the biggest changes to the road will be self-driving trucks in so-called "platoon" formations.
Another potential environmental benefit lies in rethinking car design. If driverless cars become substantially safer than those driven by humans, automotive designers could dramatically improve upon a mechanical body whose shape and size is the compounded result of a century's worth of incremental improvements and creeping crash-safety requirements. As accident rates drop significantly, driverless cars could be lighter and smaller, and therefore more fuel efficient.
Related
- Driverless Cars: When the Internet Takes the WheelDriverless Cars: When the Internet Takes the Wheel
Driverless taxis would not be the only vehicles to shrink in size. The delivery of packages and food orders could be handled by tiny, lightweight autonomous delivery drones on wheels. On college campuses, pizza, the perennial U.S. favorite, would be delivered in plastic, wheeled autonomous "pizza drones," baked to just the right consistency during the ten-minute journey. Contrast that with the nearly one-ton vehicle required to deliver a one-pound pizza today. Most of that ton of weight is for the benefit of the human driver, not for the pizza.
One core characteristic of cars that could be improved upon is how they're powered. Driverless cars will likely have electric engines. One of the barriers to the adoption of electric-only cars has been a lack of widely available methods for charging the car's battery. Tesla has overcome this limitation by building its own recharging infrastructure. As cars become intelligent enough to plan their journeys to include pit stops at charging stations, much of the uncertainty associated with an engine that needs regular recharging will be reduced.
A combination of energy-saving benefits, including platooning, lightweight car bodies, efficient driving, and rechargeable batteries will minimize some of the negative effects of driverless cars. Another environmentally degrading activity that most of us participate in on a daily basis is parking. Driverless cars will improve city life by reducing cruising, the tedious circling that drivers do when in search of a parking space, and by doing away with the need for parking lots altogether.