The company’s vehicles saw an 85% reduction or 6.8 times lower crash rate involving any injury from minor to severe and fatal crashes or 0.41 incidence per 1 million miles for the Waymo Driver versus 2.78 for the human benchmark, Waymo’s blog said.
The study also found a 57% reduction or 2.3 times lower police-reported crash rate or 2.1 incidences per 1 million miles for the Waymo Drive versus 4.85 for the human benchmark, the blog said.
“This means that over the 7.1 million miles Waymo drove, there were an estimated 17 fewer injuries and 20 fewer police-reported crashes compared to if human drivers with the benchmark crash rate would have driven the same distance in the areas we operate,” the blog said.
The study reviews the miles the vehicles have driven in the company’s active markets of Phoenix, San Francisco, and Los Angeles.
Waymo said its vehicles saw similar results when the data was localized. Phoenix saw an 80% lower crash rate involving injuries and San Francisco a 70% reduction. Los Angeles didn’t have enough drivable hours to compare.
Waymo made adjustments to the data to correct what it says are biases in data collection. This includes underreporting from human crashes.
“While the data on human crashes that lead to injuries or property damage is fairly robust, a large number of human low-severity crashes — like hitting some road debris or minor ‘fender benders’ — are not reported to police,” Waymo’s blog said. “In contrast, AV companies report even the most minor crashes in order to demonstrate the trustworthiness of autonomous driving on public roads. For example, only 21% of crashes that Waymo has reported to NHTSA to date have resulted in a filed police report, regardless of the party at fault.”
“Not-in-transport” or parked vehicles involved in a crash were removed from the findings. Most crashes that happened at a speed less than 1 mph were excluded from the group after review by researchers. One low-speed crash, that involved a cyclist was included in the study.
The study says it is one of the first to compare overall crashed vehicle rates using rider-only data. Previous data has focused on automated driving systems (ADS) testing with a human behind the wheel, it said.
Data for the study was derived from the National Highway Traffic Safety Administration Standing General Order (SGO) database. The order requires reporting crashes that involve vehicles equipped with ADS or SAE level 2 advanced driver assistance systems (ADAS).
Waymo rolled out their autonomous cars without test drivers in all three markets in the past year.
The study lists three crashes that resulted in injuries and matches three crashes with minor injuries listed in the SGO database. They happened in July, August, and October.
- In July, a Waymo vehicle was traveling westbound in the left lane of a roadway in Tempe, Arizona, when it detected a large, fallen tree branch in part of its lane. The Waymo vehicle slowed to 3 mph and was struck by a passenger vehicle behind it. The passenger vehicle was struck by a third vehicle.
- In August, a Waymo vehicle started to move through a traffic light after it turned green. The Waymo vehicle then slowed to a stop and an SUV traveling closely behind the Waymo rear-ended the vehicle. The SGO summary does not say why the Waymo slowed to a stop. The summary states a person in the Waymo vehicle reported injuries.
- In October, a Waymo vehicle was traveling northbound when it detected a passenger vehicle approaching from behind and traveling at a higher rate of speed. The Waymo vehicle briefly accelerated. Yet, it was struck from behind.
Two crashes not listed on Waymo’s table mention injury in the SGO report. A November crash with serious injuries happened outside of the timeline for the study, which extended through October.
- In August, a Waymo vehicle was struck from behind while stopped in traffic at a red light for a pedestrian crosswalk in Tempe, Arizona. The crash is labeled as having unknown injuries in the SGO report but the summary states one individual was transported from the scene to a hospital.
- In November, a Waymo vehicle and another passenger vehicle were struck by a third passenger vehicle that ran a red light. After impact with the Waymo vehicle, the third vehicle struck pedestrians on a nearby sidewalk. All three vehicles were towed from the scene and at least one individual involved was transported to a hospital, according to the SGO summary.
A second recently published paper by Waymo outlines methods the company thinks researchers should use when attempting to compare autonomous vehicle crashes to human driver crashes.
“Determining valid and comparable human crash benchmarks is key for understanding autonomous driving technology’s performance,” Waymo’s blog said. “Despite public accessibility of both human and autonomous vehicle (AV) crash data, comparing the two comes with its share of challenges. Our benchmarking paper aims to ensure a fair comparison between AV and human driving by addressing the most common errors and biases and establishing valid benchmarks from the cities in which Waymo operates.”
According to the paper, researchers should keep in mind the underreporting of human crashes, compared to similar types of vehicles such as passenger vehicles, and determine a severity level. Other factors could include geographic region and road type.
“This body of work aims to contribute to the ability of the community — researchers, regulators, industry, and experts – to reach consensus on how to estimate accurate benchmarks,” the paper said.
The paper comes as federal and state regulators continue to wrestle with how to provide oversight over the autonomous car industry.
Earlier this month, the University of Michigan released a white paper “Regulatory Frameworks for Smart Mobility: Current U.S. Regulation of Connected and Automated Vehicles And The Road Ahead.”
“With increased innovation in and adoption of connected and automated vehicle (CAV) technologies, the U.S. federal government and state governments across the country are grappling with how to responsibly regulate these new technologies,” the paper said. “Questions about CAV regulation ranging from uncertainty about how to allocate regulatory authority between federal and state governments (and between regulatory agencies) to debates over which specific safety standards should apply loom for automakers, insurers, technology companies, and other industry actors.”
Federal legislation for oversight has yet to be passed. However, multiple federal agencies, such as NHTSA and the U.S. Department of Transportation, have written their own frameworks.
“As the primary federal actor in this space, USDOT envisions a federal-state division of responsibility in which NHTSA regulates ‘the safety design and performance aspects of motor vehicles and motor vehicle equipment’ while states regulate ‘the human driver and vehicle operations,’” the paper said. “The Department ‘strongly urges’ states to leave safety design and performance aspects solely to NHTSA.”
Multiple states have created regulatory frameworks with a majority of attention on testing, licensing, and operating of the vehicles, the paper said. The approach has varied depending on the state.
California, for example, has created a separate testing program for autonomous ride-sharing. Waymo and General Motors’ Cruise were the first companies to receive permits for autonomous passenger service with a safety driver in March 2022. The companies later received permits for driverless testing permits.
The California Department of Motor Vehicles (DMV) suspended the deployment and driverless testing of Cruise in October, in part due to safety, but also because the state said Cruise “misrepresented” information related to the vehicles’ safety.
The suspension came days after safety concerns regarding an August collision between a Cruise vehicle and a firetruck in San Francisco as well as a pedestrian-involved incident in October.
The University of Michigan paper said state and federal agencies are continually updating AV regulations.
“Bottom line, state and federal governments are far from finished in their efforts to create workable frameworks that encourage advancement and deployment of CAVs and CAV technologies, and more activity is bound to occur,” the paper states. “Stakeholders seeking to innovate in this space should watch these developments closely, consider how best to implement or take advantage of emerging programs, regulations, guidance, and standards, and look for opportunities to productively engage with regulators and lawmakers and help develop a record that will inform future frameworks to advance the development and deployment of CAVs and CAV technologies.”
Waymo addressed the need for collaboration between autonomous vehicle companies and lawmakers in a paper released in March. Proper safety research, as discussed in Waymo’s recent reports, was also partly the focus of the March paper.
“We believe the safety case-based approach can provide regulators with an unprecedented line of sight into the basis for each company’s determination that its system is ready for deployment,” Waymo said. “Yet, in order to develop a worthwhile safety case, it is first important to understand what makes one credible and well crafted, and align on evaluation criteria.”
Photos courtesy of Sunday Photography/iStock