The Fuse

Study: Put Autonomous Vehicles on Roads Sooner Rather Than Later

by Matt Piotrowski | November 15, 2017

Autonomous vehicles (AVs) should be deployed once they reach the important milestone of being safer than human drivers, according to a new report from think tank RAND. The authors argue that introducing AVs “sooner rather than later,” instead of waiting for the technology to be “perfected,” would save more lives over time from a reduction in traffic deaths. Faster deployment would lead to quicker enhancements in AV safety. “The more miles that autonomous vehicles travel—on different roads, in different environments, and under various weather conditions—the more quickly their safety improves,” the report said.

Introducing AVs “sooner rather than later,” instead of waiting for the technology to be “perfected,” would save more lives over time.

Securing America’s Future Energy (SAFE) Commission on Autonomous Vehicle Testing and Safety made a similar recommendation earlier this year, arguing that AVs exceeding human drivers’ performance would provide a “net benefit to society.” It added: “Such a standard for AVs represents a public commitment to ensuring a reasonable level of safety performance during developmental stages, which will help foster public acceptance.”

If you look at the chart below, it shows the effects of AV penetration starting in 2020—when they are only marginally safer than having a person behind the wheel. With “better-than-average” self-driving cars introduced early next decade, AVs would eventually lead to a sharp drop in vehicle deaths by the mid-2030s, the point at which AVs near perfection. From that point until 2070, some 1.1 million lives would be saved by autonomous technology.


Waiting for perfection has its downsides. The longer the country delays the widespread penetration of AVs, the more lives are at risk. RAND estimates that if AVs are not introduced until 2040 (and therefore not perfected until 2050), the new technology will reduce deaths by 580,000 over the five-decade period—roughly 500,000 lower than the number saved from a rollout in 2020.


“It might sound counterintuitive that waiting for safer cars would save fewer lives,” said authors Nidhi Kalra and David Groves in the report. “But the most important factor is time.” SAFE’s Commission made a similar argument: “Holding this new technology to an unreasonable safety standard may delay deployment, mass adoption, and the resultant safety benefits.”

Convincing the public of benefits of AVs

The challenge for policymakers and industry is winning over a skeptical public and convincing people that AVs are safer than the average driver and that they are an overall positive for society. At what point AVs will become “perfected” and cause virtually zero accidents is still to be determined. Given the infinite number of scenarios a car could encounter on the road, it will take years or decades before AVs are finally flawless.

In the meantime, drivers and passengers will need to be persuaded through testing, demonstrations, data, and personal experience that AVs are safer than vehicles with humans behind the wheel. Last year, more than 37,000 died on the country’s roads, with drinking and driving and distracted driving as major reasons for the high number. AVs can address these failures. But some drivers and passengers may not trust the technology or they simply will want to continue to drive themselves. The worries are expected to subside over time, and attitudes toward autonomy will likely shift. Government regulators, policymakers, and industry understand that it will take a long-term commitment to change public perceptions.

As the RAND study points out, other issues besides safety have to be dealt with before AVs hit the road at a commercial scale. These include insurance, privacy, cybersecurity, and infrastructure. Discussion over when AVs should be deployed comes at the same time policymakers on Capitol Hill have advanced AV legislation through both chambers. The bills that overwhelmingly passed the House and the Senate include federal preemption provisions to prevent a patchwork of state-by-state regulations. And they would give automakers exemptions from safety standards, allowing companies to field AVs whose features do not fit under current regulations and accelerate adoption.

Besides the safety benefits of rolling them out as soon as possible, increased fuel efficiency and fleet diversity, access to transportation for the disabled and elderly, and reduced congestion can all be realized through autonomy.

Testing is accelerating

The necessary testing by stakeholders is now occurring at a fast pace. A number of companies, including GM, Uber, nuTonomy, and Waymo, have launched on-road testing of AVs. In fact, Waymo is now the first to test self-driving cars without a safety driver behind the wheel. Trials are taking place in a defined space on public roads in Arizona, and Waymo’s fully autonomous cars are sharing the roads with conventional vehicles and pedestrians.

Besides on-road tests, companies are also using simulations and closed-track testing to gather data, with Waymo’s fake city in California, the American Center for Mobility in southeast Michigan, the University of Michigan, among others as key testing areas. The more testing and data companies can gather, the more opportunities they will have to establish metrics to determine when AVs are safer than humans. Then once AVs are deployed at a large scale, they can be perfected at a faster pace—and ultimately save more lives.