Waymo has released new research saying that its driverless robotaxis reduce pedestrian- and cyclist-involved collisions by 82%-92%, and crashes that involve an injury by 96%, when compared to the average driver.

Waymo has been operating its autonomous, driverless Level 4 robotaxis for several years now, and is continuing to (slowly) roll them out to more metro areas in the US. They’ve been operating in Phoenix since 2019 in some capacity, and entered San Francisco in 2022, Los Angeles in 2023, and Austin, Texas in 2024, plus they’ve just started testing in Atlanta, Georgia.

In that time, the company has racked up 56.7 million miles of operation, allowing it to have a big enough sample to start understanding how its driving capabilities compare to the overall vehicle fleet.

Today it released a research paper that it has published, suggesting that its vehicles are indeed quite a lot safer, especially when it comes to “vulnerable road users” like pedestrians, cyclists and motorcyclists.

Waymo released a table showing the total number of incidents it had in each location, inclusive of not just crashes with vulnerable road users, but vehicle-to-vehicle crashes as well.

But in its press release it highlighted vulnerable road users specifically, showing that Waymo’s robotaxis had a 92% reduction in crashes involving an injury with pedestrians and an 82% reduction with both cyclists and motorcyclists. This increases to a 96% reduction in injuries in intersections, which are one of the most dangerous parts of the road, and 85% reduction in crashes with “suspected serious injuries” or worse.

Due to the high number of miles studied, results for some specific environments and types of crashes are statistically significant, but some of Waymo’s other results are not – because some of these types of crashes are extremely rare. So more research will come as more miles get racked up.

Along with its blog post, Waymo released a short video with some examples of avoided crashes with vulnerable road users:

Waymo’s results show a particularly stark difference given that pedestrian injuries are at a 40-year high in America. Until around 2012, the trendline for pedestrian injury and death was trending downwards, showing that roads and cars were getting safer for other road users.

But in the last 13 years, pedestrian deaths have shot up rapidly, owing both to distracted driving (cellphones, screens in cars) and much larger vehicles. The latter point is finally seeing some attention by the US government – or at least, it was, before this January rolled around (we’ll see if republicans continue on their quest to make literally everything worse).

Waymo’s research has been accepted for publication in the scientific journal Traffic Injury Prevention.

Electrek’s Take

We took a ride in a Waymo when the service first came to LA and you can read my long writeup of that here, including lots of video showing how the car performed in some pretty difficult road situations. I was quite impressed, but it still isn’t perfect.

But Waymo has put quite a premium on safety, which it can do because it’s funded by Google’s deep pockets. It has spent quite a bit of money on developing and attaching its sensor suite to its robotaxis, and the statistics seem to suggest that that expenditure has paid off.

Though it sort of already has paid off, as Waymo’s main driverless competitor, Cruise, ended operations in 2023 after a high-profile crash. Cruise’s vehicle was not at fault for the crash (a human driver caused it, hitting a pedestrian into the Cruise vehicle), but Cruise subsequently was found to have misled investigators, which was a big no-no.

Waymo’s sensor-heavy is different than the approach taken by another company that talks a lot about self-driving, Tesla. Tesla is using a camera-only system, whereas Waymo has several other sensors, like radar and LiDAR (you may have heard about the difference between these two in a recent controversial Mark Rober video, which everyone seems to have missed the point of).

There are some strengths and weaknesses of each approach, and time will tell which one works out the best. Tesla’s solution is more scalable and the company has far more road miles covered than Waymo does, but the quality of Tesla’s data is lower due to its smaller (and cheaper) sensor suite.

Tesla occasionally releases a safety report, but the data included is quite minimal and has not been published in any scientific journals for peer review.

But most observers (other than Tesla CEO Elon Musk, whose observational capacities are questionable these days; and Andrej Karpathy, a well-respected top AI researcher and former Tesla AI lead) think that camera-only is not going to be able to get us to true self-driving vehicles.

Charge your electric vehicle at home using rooftop solar panels. Find a reliable and competitively priced solar installer near you on EnergySage, for free. They have pre-vetted installers competing for your business, ensuring high-quality solutions and 20-30% savings. It’s free, with no sales calls until you choose an installer. Compare personalized solar quotes online and receive guidance from unbiased Energy Advisers. Get started here. – ad*