Prototype from Google's self-driving car project. Image by Marc van der Chijs licensed under Creative Commons.

I have no doubt that fleets of autonomous vehicles, dispatched via computer algorithms similar to those coordinating today’s app-based rides, will someday roam DC’s streets and substantially impact how we get around.

That could be a very good thing. Improvements to traffic safety resulting from the transition from human-operated to autonomous vehicles could save tens of thousands of American lives per year. Their ability to travel independently to transport other fare-paying passengers will allow area residents to re-allocate the tens of thousands of dollars they currently use to own, maintain, and store cars to other, more fulfilling endeavors.

However, some people envision these vehicles as totally usurping mass transit–a misguided ambition which will instead lead to chaos and congestion on our streets. In fact, we’re seeing early echos of this future now.

What does a future with autonomous vehicles look like?

There are two diverging visions for autonomous vehicles.

In the first, private autonomous vehicles serve as the backbone of our transportation network, a scenario that the Cato Institute, a libertarian policy think tank, advocates for:

“Congress should stop funding expensive and obsolete rail transit projects, which will have no place in a future likely to be characterized by widespread sharing of self-driving cars…[and Congress] should end the mandate for states and metropolitan planning organizations to write long-range transportation plans, as planners cannot predict the effects of autonomous vehicles.”

In the second, these vehicles serve as one useful component of a multimodal system, which is the “three-revolution” scenario presented in a joint study from the UC Davis Institute of Transportation Studies-Institute for Transportation and Development Policy. Researchers envision networks of shared autonomous vehicles that sync with and complement mass transit, instead of totally usurping it.

Since autonomous vehicles aren’t in widespread use yet, there is limited information available to understand how to maximize their benefit to public mobility. So how can we decide which vision is best?

Many experts see app-based ride hailing (such as Uber and Lyft) as a precursor to autonomous vehicle networks. That means we can gain a sense of how automated vehicles would work (or not) by studying what happens when these ride-hailing services are the main transit option available in a given situation.

DC is already seeing what it’s like when autonomous vehicles dominate–and it’s not ideal

We do not need to travel far to find such a situation. Late at night, these apps dominate the scene in neighborhoods such as U Street and Adams Morgan when Metrorail is not running, bus service is limited, and late night revelers may not find cycling safe.

According to the Metrobus driver operating a late-night 96 I rode a few weeks back, travel through DC nightlife hotspots at 2 am is slower and more stressful to navigate than even rush-hour traffic congestion. Motionless cars block travel lanes as they wait for riders, delaying the few buses that are running. The unpredictable stops, swerves, and door openings needed for cars and passengers to rendezvous create a chaotic and dangerous situation for pedestrians, cyclists, and drivers alike.

Getting around U Street at late hours on the weekend can be a hassle.  Image by Mike Maguire licensed under Creative Commons.

In the future Cato envisions, bar closing time in DC would serve as the model for all mobility, at all hours. With no buses or trains available, lines of empty, idling autonomous vehicles would fill most of DC’s street grid at the end of each workday, waiting to transport parents home to their children, lobbyists to networking events, and baseball fans to Nationals Park. Once riders manage to locate and enter the vehicles they hailed, they would have to wait for everyone on their travel route to do the same before proceeding to their destinations.

Anyone who has experienced rush hour in a major city in the developing world knows that, without sufficient funding to construct modern transit infrastructure, people are dependent on the “jitneys and other small vehicles going from many origins to many destinations” that Cato researchers advocate for. The result is clogged roads and snarls of traffic, which have sparked a public health crisis of injuries and deaths from road traffic.

There is a reason almost all American transit providers choose to operate large vehicle-based fixed-route systems in the cities and towns they serve. Remember, transit providers have offered paratransit service and covered lightly populated areas using small-vehicle “demand response” systems, decades before Uber founder Travis Kalanick became a household name.

It doesn’t have to be this way. The UC Davis study proposes a better future.

Fortunately, we still have plenty of time to learn from these observations. We must plan for a multimodal future that properly integrates autonomous vehicles and other technological improvements into our infrastructure–not losing sight of the goal of getting people where they want to go as reliably, efficiently, and safely as possible.

The UC Davis study projects that shared, multimodal, mobility-incorporating autonomous vehicle technology could improve urban livability. This technology could also reduce the overall costs of constructing, operating, and maintaining transportation infrastructure by 40 percent, saving cities worldwide $5 trillion annually by 2050.

To achieve these cost savings and other benefits, the study suggests a combination of continued investment in trunk rail and Bus Rapid Transit (BRT) lines, improvements to pedestrian and cycling infrastructure, and growth of small (8-16 seat) bus networks.

A multimodal future for transit is best.  Image by BeyondDC licensed under Creative Commons.

High-capacity rail and BRT corridors offering fast, frequent, and reliable service would facilitate access to employment centers, entertainment districts, and other activity hubs. Residents of metropolitan areas could choose to live in dense mixed-use development projects near those corridors, or could opt for a quieter suburban lifestyle and still easily access transportation corridors using buses, shared autonomous vans, single-occupancy autonomous vehicles, or bicycles.

Some autonomous vehicles could be designed for specialized purposes. For example, shared vans on routes serving schools could have workspaces that students could use to finish up their homework or cram for tests. Vehicles serving commercial retail areas could offer extra storage for large shops. All of these vehicles could sync with rail and bus schedules to ensure transfers between modes are virtually wait-free.

We can do better

For some reason, a disturbingly high number of our region’s residents seem convinced that the future will more closely resemble Cato’s transit-free vision than a multimodal one, even though it would make them worse off.

Make no mistake: If we decide that private autonomous vehicles should be the backbone of our transit systems, we would be advocating for their failure. That failure would avert the associated reduction in personal auto ownership and oil consumption, which could adversely affect the bottom lines of corporations invested in these industries (some of whom also fund Cato.)

Even a recent Washington Post article seemed to suggest that in 25 years, Metro expansion will become unnecessary due to autonomous vehicles’ rollout. (They also assume somehow, despite the fact that dedicating a couple hundred feet of a travel lane to buses can take years of fighting, we will have constructed a top-notch BRT system by that time!)

Let’s ensure that autonomous vehicles and other technological improvements actually make it easier for us to get where we want to go. If we properly incorporate them into a multimodal future without neglecting bike infrastructure and mass transit, they certainly will.