Self-Driving Car Safety Data

Nice work by Joyce Jiayu Chen and Steven Shladover on San Francisco robotaxi crash data. Cruise and Waymo compared to Uber TNC. Short version: Driverless Waymo is about on par with Uber crash rates. Cruise crash rates are significantly higher than for Uber drivers. Safety supervised crash rate on par with Uber drivers (which is good news, and suggests that Cruise did indeed pull their safety drivers out too soon). Important: they conclude not enough data to determine safety yet, which is entirely correct. And to be sure there is a lot more to safety than low/medium severity crash rates. Nonetheless, it is important to be looking at how things going in an objective independent way, and we should all be grateful that they have spent the time doing this. And of course there are potential threats to validity -- clearly listed on page 16. Uber drivers are likely under-reporting crashes. Whether under-reporting by a factor of 2 or 3 depends on how much weight you give the two Cruise-sponsored studies, which have significant methodology concerns discussed in this paper. TNC drivers are a notoriously higher risk driver population, so the baseline is favorable to robotaxis. Time of day differences that favor robotaxis, etc. But it is a start, and the authors are transparent about these issues. Initial Indications of Safety of Driverless Automated Driving Systems Jiayu Joyce Chen, Steven Shladover, University of California, Berkeley TRBAM-24-05951 Paper source: https://lnkd.in/e8wryTFD (Apparently you need to be logged in as an attendee to download the paper.)

Safety of autonomous vs human drivers 🧐 “autonomous service is significantly safer towards other road users than human drivers are, as measured via collision causation”: 🔹 “the Waymo Driver incurred zero bodily injury claims in comparison with the human driver baseline of 1.11 claims per million miles (cpmm)” 🔹 “The Waymo Driver also significantly reduced property damage claims to 0.78 cpmm in comparison with the human driver baseline of 3.26 cpm” “The result is determined by comparing Waymo’s third party liability insurance claims data with mileage- and zip-code-calibrated Swiss Re (human driver) private passenger vehicle baselines” Great study from the @swiss re friends 👉🏻 https://lnkd.in/daSdVS5Z Being in Montecarlo for the 65th edition of the Rendez-Vous de Septembre, it is awesome to observe the contribute of reinsurers to the evolution of our sector 🙏🏼

With new research, the safety impact of urban AVs has come into sharper focus. We partnered with General Motors, University of Michigan Transportation Research Institute, and Virginia Tech Transportation Institute to release this groundbreaking safety study which establishes a benchmark estimate for human ridehail driving in a low-speed, dense urban environment. In comparison, Cruise AVs were involved in: 65% fewer collisions overall, 94% fewer collisions as the primary contributor, and 74% fewer collisions with meaningful risk of injury. See the full blog here: https://lnkd.in/gMP5t93G

I recently took the Waymo ride in San Francisco, and as impressive as the ride itself was, here are some even more impressive numbers to crunch! Alphabet Inc. Google’s parent company, retains 92.5% ownership in Waymo In the real world, Waymo's cars have clocked over 20 million miles together, the equivalent of driving to the Moon and back 40 times. A quick back-of-the-envelope calculation reveals that this translates to over 40 petabytes of data generated by the cars' sensors. Each day, as many as 25,000 virtual Waymo autonomous vehicles drive up to 10 million miles in simulation, which is like driving to the Moon and back 15 times. The company operates about 300 vehicles at night and 100 during the day in San Francisco. It's interesting to note more demand at night versus day; this could be due to: 1. Less availability of regular taxi service during the night. 2. Safety - passengers at night feel safer with no driver versus an unknown driver. The claimed accident rate is an 85% reduction or 6.8 times lower crash rate involving any injury, from minor to severe and fatal cases (0.41 incidence per million miles for the Waymo Driver vs 2.78 for the human benchmark). Waymo has partnered with Stellantis, Mercedes-Benz AG, JLR, and Volvo Cars - we may see a shift towards vehicles specifically designed for autonomous driving, offering passengers a new level of safety, comfort and convenience. Automobile innovation has come a long way, especially in the recent two decades. It is very interesting to see where the industry could be over the next two decades and its implications on jobs, the economy, transportation, and mobility. Commercial logistics and large fleet operators could change the dynamics if and when they go autonomous, on roads, rails, and water - if not in the air, yet. What do you think may happen in the years ahead?