The Minnesota Department of Transportation (MnDOT) released a new 20-year plan in December 2013 – the Minnesota State Highway Investment Plan (MnSHIP). Spoiler alert: MnSHIP says we need more money to build more road capacity.
MnSHIP says much more than that, but adding road capacity is a central theme, as has been the case for many years with such long-range plans. At first blush, the call for increased road capacity seems like the most unassailable part of the plan. After all, Minnesota’s population is expected to increase over the next 20 years.
But the call for additional road capacity could ultimately turn out to be the most flawed part of the plan. Here’s why: One term you won’t see in MnSHIP is “driverless car.”
Are Driverless Cars Feasible?
When I was a lad, the science fiction cartoon The Jetsons offered the dream of flying personal vehicles, which, alas, have not materialized. That has made many of us skeptical about subsequent predictions about revolutionary transportation technology, such as Personal Rapid Transit (PRT) and driverless vehicles.
But driverless cars are far less speculative that flying cars. Google, one of the wealthiest corporations on the planet, has been investing heavily in a driverless car. Their test vehicle has logged over half a million miles, and it has never had an accident while the computer was driving.
Based on their tests to date, Google founder Sergey Brin predicts that Google will have autonomous cars available for the general public by 2017. Again, this isn’t some penniless, garage-based tinkerer expressing his utopian pipe dream. This is the founder of a company bringing in almost $15 billion in revenue per year. This is someone who has already produced a prototype that is successfully operating on the streets and has been legalized for use in California, Florida and Nevada.
Beyond Google, just about every major auto manufacturer is engaged in developing this technology. If Google doesn’t nail the driverless car assignment, one of their well-resourced and experienced competitors might.
Ignoring driverless cars in a 20-year transportation plan beginning in 2013 plan may turn out to be akin to ignoring horseless carriages in a 20-year transportation plan written in 1903. Ford Motor Company was founded in 1903, and by 1923 Ford was flooding 2,000,000 Model T cars per year onto an overwhelmed infrastructure.
But will consumers really surrender control of their vehicles to a computer? In 2017, the first year Google predicts that driverless cars will be available to consumers, we won’t see mass consumer buy-in. It will take time for the skeptical masses to observe the early adaptors. But within the 20-year sweep of the MnSHIP era, broader consumer buy-in is certainly a distinct possibility.
Safety Advantages. Driverless vehicles could offer consumers significant advantages. Any life insurance underwriter can tell you that driving is one of the most dangerous tasks any of us regularly undertake, and driverless vehicles offer the hope of vastly improved safety. Though human egos makes us skeptical of this truth, computer drivers have the capacity to be much more attentive, reliable and quick to react to danger than even the most skilled human drivers. In this way, the computers have the potential to keep us safer than human drivers can.
Time-Saving Advantages. Driverless cars also can offer us more of life’s most precious and limited commodity — time. Distance sensors and computers allow computer-driven vehicles to safely follow each other at much closer distances and higher speeds than human-driven cars, making for shorter, less congested and less stressful trips.
If driverless cars can supply Americans with more time, less stress, lower insurance rates, and less death and suffering, consumers will demand it. If policymakers further stimulate such consumer demand with incentives, such as tax breaks or dedicated lanes that offer faster and safer service, the revolution could happen even more quickly.
At first blush, the dawn of the driverless car era doesn’t seem to have implications for a transportation plan like MnSHIP. After all, we would still need roads for those driverless vehicles, right?
While we would still need roads in the era of driverless cars, we might need much less road capacity, and different kinds of road capacity.
Both because of fewer crashes and vehicles that can follow each other more closely at higher speeds, we might need much less road capacity to serve travel demand. How much less? Patcharinee Tientrakool of Columbia University estimates that autonomous vehicles could improve capacity by 43%. Driverless vehicles that can coordinate with other driverless vehicles would increase capacity by 273%.
Adeel Lari, a transportation expert and former MnDOT leader who is now at the University of Minnesota’s Humphrey School of Public Affairs, points out that in the 1960s traffic engineers were taught that highway capacity maxed out at around 2,000 vehicles per lane per hour. With improved traffic management methods and technology, Lari and his MnDOT colleagues later found they could briefly push capacity as high as 2,600 vehicles per lane per hour.
Temporarily moving from 2,000 to 2,600 vehicles per lane per hour was a huge improvement. But driverless cars could push capacity to a jaw-dropping 6,000 vehicles per lane per hour or higher, which Lari calls a “game changer.”
Transportation Planning Implications
For MnDOT, here’s how “the game” could rapidly change:
- Less Road Capacity? Minnesotans might need much less road capacity at a much lower taxpayer cost.
- Narrower Lanes? We also might be able to use narrower lanes, since driverless cars could reliably navigate tight spaces, and squeeze more vehicles through choke points in the process.
- Dedicated Lanes? In the interim period when both human drivers and computer drivers are sharing the roads, it might make sense to have dedicated lanes for driverless vehicles, to keep them safe from more erratic and less skilled human drivers.
- More sprawl? If driverless cars allow for shorter and less stressful trips, people may feel free to move further away from their jobs and other destinations. If they do, the increased sprawl would impact infrastructure needs.
- Gas Tax Alternative? Safer driverless vehicles might be able to be much lighter, and therefore be more fuel efficient. Additionally, less stop-and-go traffic would also save fuel. While these changes would be good for the environment and energy security, they also would mean less gas tax revenue available for maintaining and retrofitting the transportation infrastructure.
These are just a few of the kinds of issues transportation leaders should be analyzing. Land use planners have their own set of issues to analyze.
MnDOT, and its MnSHIP collaborators at the Met Council, are wise not to construct MnSHIP on an assumption that mass use of driverless vehicles is imminent. I’m not naive about all the variables that could delay or stop the successful development and deployment of this technology, or the public acceptance of it.
But in a 20-year plan, it is an oversight to ignore the potential implications of an issue as distinctly possible as driverless vehicles. MnSHIP should call on MnDOT and Met Council leaders to closely monitor and analyze the pace of driverless vehicle development, and consumer buy-in, so they could, if necessary, swiftly adjust their plans to fit a newly emerging reality. After all, the transition to driverless cars will be no time for vision-less planning.
I don’t think you’re wrong, but I am afraid that driverless cars will be a high-tech distraction from the real reason we shouldn’t be building more roads, which is that we should be building transit, building density, and encouraging biking. It will be an excuse to avoid reforming the way we build cities and towns, and that would really be a tradgedy.
…. er tragedy.
Google’s cars have been able to log many miles overall in automated mode, but they have required fairly frequent intervention — if they were running completely on their own, they’d probably create a bad day for their passenger every 50,000 miles as things currently stand. They’re not quite as good as humans, or perhaps on par, from a safety perspective. There’s a potential for them to be many times safer drivers than humans, but the impact on traffic congestion will probably be much less significant.
I’m really not convinced that driverless cars will help the capacity of existing roadways all that much. There will be some benefit from a “driver” that never gets angry or tired, starts and stops more smoothly, and avoids unwise maneuvers, but the impact on congestion from having well-behaved drivers will only be incremental as opposed to the relative revolution they should have for safety.
Counteracting the benefits of smoother driving will be a strong desire to use robo-cars as a single-passenger private transit service. We’re currently in a world where we have about 1.1 or 1.2 passengers per vehicle on the road (usually just the driver, but there are occasionally others in the car as well). Driverless cars will probably push that ratio below 1.0, as the cars automatically seek out cheaper (or free) parking on remote parking lots, side streets, or perhaps all the way back at the owner’s home. Take the cost to park at a nearby ramp and divide that by the cost per mile to own or operate the vehicle, and it seems reasonable that many self-driving cars would go miles out of their way to park for free.
Things may not be quite that bad if driverless cars can turn into a popular taxi-like service, or an automatic carpool, but there will still be significant periods of time when they are driving around empty.
There’s also the problem of pick-up and drop-off zones. Are downtown thoroughfares all going to turn into something like those zones at MSP airport? How good will cars be at recognizing their passengers waving at them from the sidewalk, and vice versa?
We’ll come up with some rules for how to handle that stuff, but there will be tradeoffs. Cars may be required to park close to where their owners are, or operate much like today with spaces where they park all day long, but that obliterates the possible benefit of developing downtown land more uniformly. How far can we turn the knob toward allowing remote auto-parking without gumming up surface streets at rush hour or letting arterials and freeways be clogged with the post-/pre-rush empty cars as they maneuver around?
Human drivers are also going to remain for a long time — almost certainly 20 years from now, so we can’t take out many planned upgrades currently justified by safety concerns. (Of course, we probably should try to change many dangerous intersections to roundabouts, etc., rather than using stoplights or expensive grade separations, but that holds true whether it’s a human or a computer behind the wheel.)
Robot cars for safety? Sure! Robot cars as a congestion cure-all? Not so much.
Great comments Mike. I’d like to see this expanded to a whole counterpoint post. “The Potential Drawbacks of Driverless Cars” or something like that.
Not necessarily questioning your facts, but can you point to a reference for your assertion that Google’s testing has required frequent intervention? I hadn’t heard that before.
“Left to its own devices, Thrun says, it could go only about fifty thousand miles on freeways without a major mistake. Google calls this the dog-food stage: not quite fit for human consumption. ‘The risk is too high,’ Thrun says. ‘You would never accept it.’ The car has trouble in the rain, for instance, when its lasers bounce off shiny surfaces. (The first drops call forth a small icon of a cloud onscreen and a voice warning that auto-drive will soon disengage.) It can’t tell wet concrete from dry or fresh asphalt from firm. It can’t hear a traffic cop’s whistle or follow hand signals.”
When it comes to emerging technologies, it’s a fool’s errand to predict the future over a two-year horizon, much less over the 20-year horizon this plan covers.
So I am less certain than Mike is about how it will all shake out. He may or may not be right.
But the questions Mike raises here, among many others, are exactly the kinds of questions planners should be monitoring. Doing so is especially important with a technology as potentially disruptive as this one.
Important points. I’m one of those who thinks that the impact will be bigger than most transportation agencies are planning for. I think the biggest impact will be that the kind of taxi and car services that are currently only enjoyed in Manhattan, where you can fairly reliably get a taxi within a couple of minutes of desiring one, and where the price of the trip is fairly competitive with driving, will be extended to most urban and even suburban areas in the country. You can already see the beginnings of it in the ride-sharing services like Lyft and Uber. In many places, it takes a half an hour or more to get a taxi, but Lyft/Uber can be there in five minutes. And that’s at the very beginning of the trend. If you imagine a Lyft/Uber-like service with automated cars, at least one of which can reliably be expected to be available within a block or two of any given point in an area, at a cost that subtracts labor from the equation, then you can start to imagine that a lot more people will start to rely on that as opposed to their own private cars. And the roll-out of that once technically feasible could be very fast given the profits that stand to be made by whoever provides that service.
I agree that the call for additional road capacity could ultimately turn out to be the most flawed part of MnDOT’s plan. But, like Jeff, I disagree about the “here’s why.” Regardless of whether driverless cars pan out soon, more people will use public transit, bike, walk, or avoid a physical commute altogether. In 2008, the number of cars on the road decreased for the first time since probably the dawn of the automobile. Traffic engineers, including those at MnDOT, are still projecting as if this we were living in the 1990’s. I think they have to reevaluate their models to account for recent trends showing the car ownership is shrinking rather than steadily increasing as they persist in projecting evidence be damned.
Long term, driverless cars could be a gamechanger, but I wonder, to approach 6000 vehicles/lane/hour, would all of the cars need to be driverless? Or is the idea the driverless cars are so proficient they can avoid accidents even with extreme human error on the part of the other cars?