Feb. 17, 2023
In April 2022, San Francisco police pulled over a car for driving at night without headlights. But when the officers went to talk to the driver, they discovered the car was empty.
The driverless vehicle belonged to Cruise, a subsidiary of General Motors, which is one of seven companies permitted by the state of California to test their autonomous vehicles (AVs) on streets without a driver.
While the officers were considering what to do, the car started up, crossed an intersection, and came to rest on the other side. Videos of the incident, which can be seen on various news sites, show the police talking on their phones and looking baffled. They eventually get back in their cruiser and drive away.
The car had been programmed to recognize when it had been stopped by police. However, it had also been programmed to only remain stopped at places that met specific criteria, which is why it moved across the intersection without warning. The car had evidently not been programmed to prevent it operating at night without headlights.
San Francisco can ticket driverless vehicles that are parked illegally, but citations for a moving violation are given to a driver rather than a vehicle. Right now, there is a regulatory gap: Where there is no driver, the police can't give out a ticket.
Getting driverless vehicles to operate safely in every place and under any condition is a formidable task, and it is still a long way from being done successfully.
Local planners can plan for autonomous mobility, including redesigning streets to make it easier and safer for driverless cars to operate. And AV manufacturers can do their part too, starting by working with local governments and planners to safely integrate driverless vehicles into cities.
Full speed ahead
Across the U.S., driverless vehicles are being tested for paid ride services, transit, deliveries, and interstate trucking, and the companies doing that testing are looking for these services to begin full operations soon.
The website for TuSimple, which is working on driverless long-distance trucking, makes clear the benefits of automation: "Autonomous trucks have the capacity to operate nearly continuously, stopping only for refueling and preventive maintenance." The technology will allow TuSimple to reduce direct labor costs, improve fuel consumption, and seek better insurance premiums tied to lower accident rates.
Converting existing services to driverless vehicles is a huge potential market for vehicle manufacturers and software companies, and they are competing to put their AVs on the market as quickly as they can.
In the U.S., 29 states permit AV testing and operation, depending on computer-aided abilities to read and understand existing conditions. These tests have so far been confined to places and times when there is little traffic and the weather is favorable, but the intent is to move on to more difficult situations.
Even as companies race to design and build autonomous vehicles, there has been little communication between their computer programmers and vehicle designers and the local planners who could help them. There has also been little or no communication with local governments that control the streets and are charged with regulating traffic to ensure public safety.
So, the question is: Do states, which are issuing the permits now, really want to be involved in local issues about parking, loading, electric charging, rights of way, priority at road intersections, and signaling? These issues will all be important as driverless vehicles move out of the testing phase. Or will states eventually delegate those powers to local governments?
San Francisco takes the lead
Introducing driverless vehicles as part of the traffic mix raises many questions: How safe are they for passengers, as well as for others sharing the streets? What regulations are needed to accommodate their operations? Will these robots with four wheels eliminate jobs and pose a threat to the economy and vitality of communities?
San Francisco is seeking to answer some of those questions, or at least get out ahead of them. The city has developed 10 guiding principles for the new kinds of mobility services that are emerging in its streets and neighborhoods.
These principles are useful for framing the questions that planners working for local governments will need to consider before allowing the full operation of driverless vehicles. The state agencies permitting testing should consider them also.
1. Collaborate for effectiveness
The need for collaboration among manufacturers, operators, and governments is perhaps the most important of the ten principles. NACTO, the National Association of City Transportation Officials, has published a Blueprint for Autonomous Urbanism, detailing the kinds of changes to urban streets that could also feed back into the design of the autonomous driving systems themselves. But these changes would take years to implement and would require all parties to have talked to each other about the planning and design of streets and land uses.
When and how can local governments, manufacturers, and state regulators sit down and figure out how to address common problems together?
2. Help reduce traffic deaths
This second objective raises the central question: Are completely autonomous vehicles safer than vehicles with drivers? Many cities are implementing Vision Zero policies that seek to eliminate fatalities and injuries caused by moving vehicles.
Manufacturers claim that driverless vehicles will make fewer mistakes than drivers, but is any level of failure acceptable? If the technology is not 100 percent reliable, governments might require safety drivers, which would remove one of the big economic incentives for developing the technology.
Are AV manufacturers aware of Vision Zero improvements to streets and highways planned or already in progress? What is the safety standard for driverless vehicles — is it also zero traffic deaths?
3. Relate to public transit
Will lower labor costs for driverless car services take riders away from public transit? It could happen, but governments could counter with a congestion fee on ride services for long trips in downtown and inner suburban areas. Driverless shuttle buses between transit stops and frequent places of departure or destination will reinforce existing transit and could be exempt if there is a congestion fee.
How can AV operations be designed to support public transit systems and not undercut them?
4. Help manage traffic congestion
A study of transportation network companies (TNCs) in San Francisco shows that cruising streets between fares increases the carbon emissions generated by conventional TNCs, such as Uber or Lyft, to one and half times that of regular auto traffic. Electric vehicles will not have the same emissions problem, but they will still be driving around looking for customers, meaning that they will increase traffic congestion.
Planners should expect that many small, driverless delivery vehicles will add to the congestion problem when they park in front of buildings to make deliveries. These AVs on sidewalks could enter the mix when they cross streets. Local governments may have to consider limiting delivery hours in congested areas, limiting autonomous vehicles to streets and lanes designed to accept them, and instituting congestion pricing.
Breakdowns of autonomous ride-service vehicles are already causing traffic congestion during the testing phase in San Francisco. When these vehicles break down, a technician has to be dispatched to take charge of the car, restart it, and drive it away.
How much will driverless ride services increase congestion in existing streets? What can be done to manage this problem?
5. Help reduce emissions
Driverless vehicles are almost always electrically powered, which certainly helps reduce emissions locally. But that isn't the whole story. For decades to come, conventional vehicles with drivers will continue to occupy streets and highways, mixing with autonomous vehicles and adding to congestion. In traffic, traditional gas- and diesel-powered vehicles will still idle and create emissions. Solving these problems will require changes to street configurations, made through collaborations between governments and vehicle operators.
What changes to street configurations could help ease congestion that would otherwise be caused by driverless vehicles?
6. Provide equitable access
Driverless vehicles could make driving possible for people unable to drive a conventional car. That could open up more equitable access to destinations, if driverless vehicles really do end up making trips and deliveries less expensive, particularly in suburban and rural locations with little public transit.
They can deliver groceries and medicines; and driverless TNCs or on-demand shuttle buses can improve individual mobility. However, their acceptance depends on making these services affordable, as well as on designing the vehicle and the system to be friendly to seniors, children, and people with disabilities.
Can driverless vehicles make mobility more affordable, or do any cost-savings go to the vehicle owners while prices remain much the same?
7. Follow fair labor practices
Driverless technology has the potential to eliminate a lot of jobs, a major reason for developing the technologies in the first place. Right now, there is little discussion about this, but controversy is likely once implementation goes forward.
Regulations might still require a safety driver during a transition period or in bad weather; there will still be jobs monitoring the vehicles remotely for emergencies or system failures, as well as for servicing and repairing them. New service jobs could be generated by this emerging industry, like the professional shoppers in grocery stores who put together orders for delivery. But there will need to be advance planning to reduce negative impacts on the job market.
What plans do vehicle operators have to ease the transition when driving jobs start being replaced by AVs? Will there be programs to hire former drivers for other jobs needed to manage fleets of AVs? What else will need to be done to manage this transition?
8. Prevent negative impacts on urban infrastructure and services
Dealing with emergencies and unforeseen circumstances, like street repairs or emergency vehicles, will require collaboration between driverless services and transportation agencies on things like signal systems. Basic principles — like pedestrians having the right of way and not obstructing transit routes — require coordination, digital mapping, and feedback systems for the automated vehicle.
How will driverless vehicles deal safely with unusual situations like street repairs or confronting speeding fire trucks, ambulances, police cars, and other emergency vehicles?
9. Share relevant data
Governments need to make plans to address automated vehicle issues, and plans require data. The companies operating driverless vehicles will have an increasing amount of data about their operations that could be used to assist developing traffic forecasts as well as making planning and design decisions that would benefit everyone. Manufacturers and operators should share data related to issues of safety, system security, management of resources, and reduction of carbon emissions as they work with public agencies.
Manufacturers are motivated to withhold data from competitors, but they need to share it with local governments in order to collaborate on safety measures. How can the necessary data-sharing be managed?
10. Be completely accessible to people with disabilities
Driverless cars and shuttle buses should be designed to be as accessible for people with disabilities as they are for everyone else. The digital infrastructure to deploy driverless vehicles can let passengers request an accessible vehicle ahead of time. If that requires a special design and driver to assist for a certain percentage of the vehicles, there will need to be enough of them available so that wait times are not excessive.
How can local governments make sure that driverless cars and transit vehicles are accessible to people with disabilities, as there might not be a driver available to help them?
The knowledge that planners have about traffic management principles and their interrelationships with other planning objectives, as well as their understanding of local communities and conditions will be crucial in answering these questions. When will planners be part of managing the transition to autonomous vehicles?