The coming of autonomous vehicle technology shows great promise for eliminating traffic deaths and serious road injuries. They could also make life a lot easier, but only if they are done right.

Reducing traffic fatalities will depend on how cities use autonomous cars, and how cities design shared spaces for cyclists and pedestrians. In response to the development of driverless cars, there are concerns of conventional drivers having a hard time navigating a road with the addition of autonomous cars, resulting in an entirely new type of traffic crash.

The ultimate reason for death and serious injury from traffic crashes is that the human body can sustain only so much force, a key principle of the Safe Systems Approach, which aims to improve planning so that transport systems themselves eliminate traffic deaths and injuries that arise due to human error. However, recent guidance from the US Department of Transportation does not consider appropriate vehicle speeds for self-driving cars, even as United States Secretary of Transportation Anthony Foxx has said, “we don’t want to replace crashes with human factors with large numbers of crashes caused by systems.” Therefore, in order to reduce traffic fatalities and injuries through autonomous vehicles, regulation of a main factor—speed—will need to take central stage.

Autonomous Safety or Autonomous Speed?

For the most part, how quickly vehicles can react to incidents on the road and what speeds are safe for all road users will determine the overall level of road safety for all. Currently, many countries and cities do not have proper speed limits in place. While computers in autonomous vehicles may be infallible, the laws of physics still govern safety. Vehicles will need to be programmed to travel at safe speeds for everyone, especially in urban areas.

Consider pedestrians. If a speed limit is 60 km/h (37.3 mph) in an area that is also used by pedestrians, safety is not guaranteed. Even a computer may not be able to avoid a child running into the street to grab a ball. However, if the speed limit is 30 km/h (18.6 mph), or even lower, road death and serious injury are much less likely in the event that the autonomous vehicle cannot stop in time.

In addition to real safety, there’s perceived safety. Even if autonomous vehicles are shown to move safely at higher speeds, they may be perceived as dangerous by bicyclists and pedestrians. If we want to foster active and sustainable modes of transport, like walking and cycling in our cities, speeds need to be set at levels that feel safe.

Another example is the concept of platooning, in which autonomous vehicles can sync together, travelling as a sort of “train.” Yet if the syncing-up of a set of vehicles in traffic at 90 km/h (55.9 mph) at a distance of four meters apart is interrupted by, for example, a deer bolting out into traffic, the autonomous vehicles may not be able to stop in time, resulting in a crash and close-range pile up. Platooning is already undergoing road tests. Recent programs to test the technology, such as SARTRE in Europe, show that Platooning is possible, but not yet perfect. Volvo, a partner on the project, recommended heightening focus on emergency situations, such as obstacle avoidance or sudden braking.

A Speed Issue or a Moral Issue?

In addition to concerns over safety, some research presents a moral question that autonomous vehicle programming may need to address. If a moving vehicle encounters a situation where it must choose to either minimize the passenger’s risk or a pedestrian’s, which should it choose? The research found that people were in favor of minimizing trauma in general, but for maximizing vehicle occupant safety if they, themselves, were the occupant. A legitimate question—but in the Safe Systems Approach, this moral conundrum would be somewhat of a false dichotomy. Rather, the chance that the vehicle would have to make this choice is minimized because the entire transport system ensures that vehicles are moving at safe speeds for both pedestrians and drivers.

Driverless Cars Create Opportunity for a Better City

Safety and moral concerns aside, if done right, autonomous vehicles could help create better cities. Driverless cars require less road space, through reducing inefficiencies or the ability to maneuver within a narrow lane, which in theory could free up street space for pedestrians and cyclists. This leads to the question: What kind of cities will be shaped by an autonomous vehicle future? Ones that are safe and accessible for pedestrians and bicyclists, with high-quality public transport systems? Or the cities once envisioned by architects such as Le Corbusier and described by Jane Jacobs as dominated by the car, rather than built for the people? Just as most cities allowed the “horseless carriage” to take over streets, will cities make the same mistake with the driverless car?

As Dario Hidalgo, Director of Integrated Transport at WRI Ross Center for Sustainable Cities, noted, the vision of safer cities is something we should be aiming for in any case: characterized by “good urban design, compact and connected with diverse uses; high-quality design for pedestrians, safe bicycle paths and high-quality public transport.”  A chance to “redo the city,” may be on the horizon, as Zipcar Founder Robin Chase points out: “with criteria and priorities for how to repurpose that newly available public space: wider sidewalks, more street trees and plantings, bike lanes, street furniture.”

The good news is that it is still early, with time to get these questions right. At the heart of this issue is putting people first, rather than getting swept up into prioritizing the conventional flow of traffic.