Our co-founder Phil Koopman was quoted in an Axios article on setting realistic expectations for autonomous vehicle fatality reductions. While it is common to attribute 94% of fatalities to bad human driving, that number doesn’t actually mean that self-driving cars will be 20 times safer than humans. Combining a study by the Casualty Actuarial Society and NHTSA data on rural vs. urban fatalities, it seems likely that in the short to mid term the potential self-driving shared transit fatality reduction will be about a quarter of fatalities (perhaps 10,000 lives saved). It would be interesting to have a head-to-head study comparing that to what could be done with ADAS systems such as Automatic Emergency Braking, which can be deployed on the entire fleet much more quickly and be available in rural as well as urban areas.
The article was based on his slides from the OESA Annual Conference: https://www.slideshare.net/PhilipKoopman1/autonomous-vehicle-safety-improvement-less-hype-more-data