While OPs statement did make me laugh a bit (I've personally driven almost a million miles myself and I don't think my odds of killing someone were 1 in 4), we do have to realize that MOST miles that make up the 100 million mile stats are done at highway speeds far away from pedestrians.
If I had to do 4 million miles in SF, I just might have hit someone by now.
Americans drove 3.2 trillion miles in 2021 and 42,795 people died in traffic fatalities (both values are estimates, and likely do not include every single mile driven or fatality that occurred, depending on how you prefer to count).
So humans drive average of 74 million miles before a person gets killed, and Waymo's driven somewhere above 20M miles but probably fewer than 74M. At this point, I'd expect no self-driving car traffic fatalities, statistically, if the SDC fleet is an average driver.
In some sense Waymo and Cruise are basically waiting around for a fatality (not that they want one) so they start having a denominator that isn't zero.
Your statistic (vehicle miles versus passenger miles) is probably more relevant, but now I'm curious whether the passenger miles of driverless vehicles are more or less than the vehicle miles, in aggregate.
(Although maybe passenger miles is the correct metric -- a driverless, passengerless vehicle could hypothetically plow into an overpass at a speed that would result in a fatality, if anyone were inside; since the passengers of a vehicle are a large chunk of the people who die if that vehicle has an accident, should we car how many people don't die when vehicles are empty?)
Regardless of whatever the current fatal accident statistics are in the U.S., being able to build a robot that drives that far in traffic without killing anyone is an incredible technical accomplishment. The progress over the past ten years is simply astounding.
> Although 20 percent of people in the U.S. live in rural areas and 32 percent of the vehicle miles traveled occur in rural areas, 40 percent of crash deaths occur there.
> Compared with urban areas, crash deaths in rural areas in 2021 were less likely to occur on interstates and freeways (14 percent compared with 21 percent) and on other arterial roads (23 percent compared with 58 percent) and more likely to occur on collector roads (44 percent compared with 11 percent) and local roads (19 percent compared with 11 percent).
In other words, majority of rural traffic deaths are not on freeways.
Yes, but I still reject the idea that San Francisco is an especially fatal city to drive in.
Looking for statistics on San Francisco in particular, in 2016 the city averaged an estimated 5.6 million vehicle miles per day, and an average of 29 traffic fatalities per year. That puts it at about 1.4 fatalities per 100 million vehicle miles, smack dab in between the national averages for urban (1.2) and rural (1.7) traffic fatalities.
So 5 million vehicle miles continues to be a small sample size, with an estimated 0.07 deaths for San Francisco.
(I'm also not sure the drill down that urban fatalities happen on freeway & arterial roads and rural fatalities happen on local & collector roads helps the self-driving cars in SF; it seems like the speed-limited self-driving cars should be spending most of their time on the absurdly safe local roads in the city.)
There are only a few miles of freeway in the entire city of San Francisco, so I doubt too many of the accidents in SF are from them. Yes, driving 25mph in a modern car with crumple zones and 10 airbags is generally very safe. If you look at the statistics [1], most traffic fatalities in the city are either pedestrians or bicycles. Avoiding pedestrians and bicycles is one of the tougher challenges of making a safe AV, and San Francisco is one of the most challenging places in the country for this.
