Widespread use of self-driving cars in coming decades will increase congestion and pollution in the Washington region if governments don’t take steps to avoid such consequences, according a study commissioned by the D.C. government.
Ensuring the downsides don’t swamp the benefits will require careful and far-reaching action by policymakers and further in-depth study, according to the report.
“We know technology alone won’t solve all of these issues,” said Jeff Marootian, director of the District Department of Transportation, which released the report this month. “The study highlights the importance of putting in place polices that guide the outcomes we want to achieve.”
Given the rash of uncertainties — technological, political and behavioral — the study set out to understand how four future scenarios might play out. Engineering and infrastructure firm AECOM prepared the report, working with DC Sustainable Transportation, a group of nonprofit, environmental, business and governmental entities.
“Planning in the face of this ambiguity is challenging but must be tackled,” the authors wrote. “The potential benefits are too great to miss, and the potential risks too large to ignore.”
Self-driving firm Argo AI, working with Ford, is testing self-driving cars in the District and said it planned to launch a commercial service by the end of 2021. Uber this year announced a mapping and testing effort in the city, though it offered no public timeline for opening its self-driving vehicles to the public. Both said they envisioned starting with modest numbers before ramping up.
The District study sought to examine the implications going out decades, testing a range of provocative policies and possibilities.
In the first scenario, families across the Washington region would buy significant numbers of autonomous vehicles designed to drive on freeways, and Virginia and Maryland would create driverless-only lanes on freeways, pushing growth toward the suburbs.
In the second, large fleets of shared autonomous vehicles would roll out quickly in urban and other settings, replacing buses and off-peak Metrorail service while cutting individual car ownership.
In the third, officials across the Washington region would create high-occupancy-vehicle lanes for transit, targeting vehicles with five or 10 people in them. Off-street parking would be “slowly phased out,” and curbside parking would be converted into zones for dropping off passengers.
In the last scenario, a “regional congestion fee” would be put in place across all major roads in the Washington area to eliminate “extreme” traffic snarls. Autonomous shuttles would also move people around neighborhoods.
In each of those options, the total number of miles traveled on District roads would grow as more people use cars than under current regionwide assumptions. Autonomous vehicles would also drive empty between passenger trips, according to the study, among the things that would need to be addressed.
In 2019, vehicles rolled about 10 million miles each day on District roads. Based on existing regional projections, that is expected to increase 14 percent by 2045, the report said.
But under the assumptions in the first scenario explored in the study, with family-owned autonomous vehicles populating freeways, the total miles traversed by vehicles in the District would grow by 26 percent over that same quarter-century.
With the scenario covering fast-growing self-driving fleets, travel would grow about 85 percent. With the dedicated HOV lanes, travel would grow 43 percent. And with the regional congestion fee, “vehicle miles traveled” would increase by about a quarter, according to the study.
“Congestion will likely rise significantly, but public policy can change that,” the study concluded, pointing to moves to encourage ride sharing and other steps.
“Although overall congestion on the network could increase, especially outside of the peak periods, interventions like dedicated HOV lanes and/or congestion pricing could actually decrease the average trip times for people,” the authors wrote.
Such “congestion pricing” efforts have had success at reducing some backups but also can be unpopular, as has been the case with tolls on Interstate 66 inside the Beltway in Northern Virginia. The tolls must be paid by drivers traveling alone during morning and evening peak commuting hours and vary based on how many cars are on the road.
Given that the District does not have the same type of major highway networks as neighboring states, Marootian said, the nature of any such potential regional congestion fee remains unknown and is among the topics that need further local and regional study.
Still, the report points to the potential benefits its authors said could come from such an approach. It assumes “a regional effort that applies dynamic pricing on all congested freeways and major arterials in the region,” and at all times of the day, to cut traffic snarls and discourage autonomous vehicles from “spending a lot of time circling.”
Such a fee could shave six minutes off the average trip time in 2045, which was expected to be 31.5 minutes according to existing projections, the report said. Prioritizing HOV lanes would also shave off two minutes, while deploying large fleets of shared self-driving cars could add four minutes to average trip times, the report concluded.
Emissions would also soar without a big push to use electric vehicles, the report said. Marootian said new incentives and requirements would be effective and need further study.
Broad use of self-driving cars would have a host of benefits, the study found, including addressing housing affordability and spurring job growth by making parts of the District easier to reach.
A smaller number of cars would move more people, erasing the need for 120,000 cars to park in the city, according to the report. That could free parking spaces for other uses, including parks, but also crowd curbside drop-off spots and require real-time reservation systems or other tools.
The report warned that actions needed by cities such as the District to manage an influx of self-driving cars, including getting real-time data from car fleets, could be limited by efforts in Congress to “preempt” the power of state and local officials.
“Local governments should monitor any future legislation and ensure that their ability to collect data directly from . . . vehicles or operators continues to allow the flexibility needed to enact and enforce local policies,” the authors wrote.
Some self-driving-car developers and supporters on Capitol Hill have warned of a debilitating “patchwork” of local regulations that could impede the broad use of, and benefits from, the vehicles.