Self-driving cars are poised to change how we move in the 21st Century. They will be safer, more efficient, and more affordable. For most people living in urban areas, car ownership could become obsolete, while parking lots could be repurposed as housing, parks, or for other more active uses.
However, the implementation of this automated vehicle utopia is more challenging: a multitude of political, technological, and legal obstacles remain. Perhaps the biggest challenge toward full automation is a legal one: regulating a manufacturer’s potential liability when these vehicles will be operating in a complex urban environment. The presence of a dense concentration of non-automated users (bicyclists and pedestrians) will increase the chance of collisions.
To reach the dream of a fully automated vehicle fleet, we need to accomplish the goals of Vision Zero – eliminating road fatalities and serious injuries on our streets. By creating safer streets for all users, we can also create the right climate for self-driving vehicles to thrive and make our streets even safer and more livable.
Before exploring this point in greater detail, how do we define a self-driving car? The National Highway Transportation Safety Administration has outlined five levels of vehicle automation. Presently, most vehicles are at Level 0 – no automation. Some newer vehicles feature function-specific (Level 1) automation like braking or parking assistance, or combined function (Level 2) automation. Google is conducting ongoing tests with limited self-driving (Level 3) automation, which allows the driver to cede full control under certain conditions, but still may require occasional manual control. Level 4, which is not yet street-legal in California, provides full automation without any driver control – the ultimate vision for the “Google Car” that operates without a steering wheel or gas pedal.
Many of the disruptive possibilities of self-driving cars are unlocked by full automation at Level 4. Whereas Level 1 through Level 3 vehicles would continue the existing paradigm of personal car ownership, Level 4 eliminates the relationship between the driver and vehicle. Level 4 allows for an Uber-like system of driverless taxis that could make car ownership and parking largely obsolete while fundamentally reshaping the nature of urban mobility.
The key challenge toward achieving Level 4 automation is operating in dense, complex urban environments like San Francisco or Manhattan. In comparison to suburban environments where Google is currently testing their vehicles, San Francisco has much heavier volumes of pedestrians and bicyclists, and as such, exponentially more road users that are not automated. While self-driving cars remove the element of human error from driving, they also remove the element of human communication – the ability to resolve everyday situations through eye contact, waves, intuition, and “you go first” interactions. Self-driving cars may develop pinpoint accuracy in most areas, but there could still be uncertain situations in some chaotic environments: how will self-driving cars gauge the intent of dozens of pedestrians and bicyclists? And what happens if a self-driving car makes a mistake?
Liability issues of self-driving cars are largely governed by product liability law. In essence, the manufacturer will be responsible for their technology to function properly, and will be responsible for collisions resulting from breach of warranty, negligence, design defects, or manufacturing defects (each of which already have legal definitions that can adapt to changing technology).
While the legal concepts governing self-driving cars might seem clear-cut, a gray area emerges around how to define “good,” decisions for vehicles. Should a driverless car protect its occupants, even if it puts non-passengers at risk? What if those non-passengers are bicyclists or pedestrians? What legal rights will bicyclists or pedestrians have in such situations if the car did what it was “supposed” to in order to harm the fewest people? This gray area is what presents a legal risk to manufacturers.
Manufacturers and owners/operators will be incentivized to operate as safely as possible – and as such, will approach automation conservatively to limit their liability and exposure to risk. At a basic level, this approach could come in the form of extensive warning labels and disclosures. But if certain areas, like Downtown San Francisco, turn out to be particularly dangerous, and incidents there result in a number of lawsuits against them, it’s possible that manufacturers may create “manual zones” in which an occupant must take control of the vehicle and assume traditional liability for vehicle operation.
Like most discussions of self-driving cars at a stage of full (Level 4) automation, the idea of “manual zones” is only speculation at this point. However, it is clear that there are real obstacles and risks when these vehicles are operating in dense urban environments full of pedestrians, bicyclists, and humans whose actions are unpredictable and not plugged into “the grid.” But there’s a solution to reducing these risks, and San Francisco is already pursuing it: Vision Zero.
Vision Zero is a road traffic safety project started in Sweden in 1997 which aims to achieve a highway system with no fatalities or serious injuries in road traffic. San Francisco’s implementation of the Vision Zero initiative will have the ancillary benefit of setting the stage for self-driving cars to operate on its streets with a much lower risk of collision. Streets that are designed to be safe and organized for all users are streets that are most attractive for full vehicle automation. In the event the “manual zone” scenario comes true, then a complete redesign of San Francisco’s most dangerous streets could serve as an appropriate mitigation to eliminate “manual zones” in the city.
Moreover, a city that’s ready for Level 4 automation would open the door to a positive feedback loop that further enhances mobility, and creates a more walkable and bikeable city. As personal vehicles are replaced with driverless taxis, the canvas of space available for pedestrian and bicycle improvements will grow exponentially as less space is needed for vehicles – what amounts to a gigantic citywide “road diet.” Street parking and vehicle lanes could be repurposed as protected bike lanes, wider sidewalks, and improved crosswalks. Traffic signals could be retimed to feature more pedestrian scrambles (stops of all vehicular traffic that allow pedestrians to cross an intersection in every direction, including diagonally, at the same time) and dedicated bicycle phases as well. The amount of space dedicated to moving and storing cars could shrink dramatically.
There are reasons to be cautious with self-driving cars. In a worst-case scenario, self-driving cars could pressure cities like San Francisco to once again adopt a car-first mentality. Self-driving cars may result in longer commutes, more driving, and greater dependence on automobiles, while exacerbating tensions between vehicle efficiency and multimodal balance. We could rebuild our cities once again around driverless cars, rather than using this technology to complement our increasingly people-oriented cities. For a variety of reasons – “manual zones,” manufacturer reluctance, technical limitations, poor planning, and others – it could take a long time to get to a self-driving utopia. Or, we might never get there.
Regardless, while it’s certainly not necessary to find more reasons to support Vision Zero, these investments could ultimately pay off in opening new opportunities for self-driving cars. Neither self-driving cars nor Vision Zero improvements will eliminate all traffic collisions in San Francisco, but together, they might come close.
The views and opinions expressed on this post are those of the original author(s). These views and opinions do not necessarily represent those of Walk San Francisco, nor does this post indicate any endorsement, explicit or implied, for the views and opinion of the author(s).