Ohio Self-driving Shuttle Service Stalled After Minor Incident

I dunno, Teslas are kinda doing a good job stopping themselves by killing their inattentive drivers.

@Corey Lewis Corey, around 166,000 people have been killed in roadway fatalities since your January 10 safety post. Do you want to snark on Tesla, or are you interested in saving lives? (If the latter, you do have some potential positive influence to wield.)

@ToolGuy Tesla-specific data: https://www.tesladeaths.com/ 33 Tesla drivers have died in Teslas in the history of the planet. 119 people globally have died in roadway fatalities where a Tesla was involved in some way - ever. Over 3,600 people are killed each day in roadway fatalities around the world. https://www.cdc.gov/injury/features/global-road-safety/index.html

@ToolGuy Those numbers are entirely meaningless, since the number of Tesla's on the road compared to all motor vehicles at 0.006% can be rounded to zero. The real important number is deaths per million km (miles)driven, and where those km are driven. The roads in Africa and parts of Asia have much higher rates of road fatalities per km than Europe or North America, yet relatively fewer Teslas. The bulk of Teslas drive on the safest roads, so naturally you'd expect their fatality level to be lower.

@ToolGuy U.S. 2018 rate was 1.13 per 100 million vehicle miles traveled. https://crashstats.nhtsa.dot.gov/Api/Public/ViewPublication/812826 Don't know the U.S. 2018 Tesla rate, but... There were 12 U.S. Tesla-involved deaths in 2018. https://www.tesladeaths.com/ At the end of 2018, there were approximately 331,521 Teslas in operation in the U.S.: http://carsalesbase.com/us-car-sales-data/tesla/

@ToolGuy 26 to 180 deaths out of 2.2 million vehicles sold for the Ford Pinto and we are still talking about it 40 years after the last one rolled off the assembly line. You don't get a pass when your vehicle kills their occupants in a preventable manner. They need to fix this.

@ToolGuy And the fact that 2.2 million people purchased a freaking Pinto over 10 years puts those Tesla numbers in a bit of perspective. Was 2011 their first year? And not all of those have the "self driving feature"...those are the deaths in question here. It is not a mathmatically insignificant ratio when you look at cars like the Corvair, Pinto, GM Crapboxes, and some of the other safety campaigns the NHTSA has gotten involved with.

@ToolGuy Back with more math, if I may: Say that the 331,521 Tesla vehicles in operation in the U.S. at the end of 2018 each did an average of 10,000 miles in 2018 (some more, some less, obviously some weren't on the road for the entire year). That's 3.3 billion vehicle miles traveled by Tesla vehicles in the U.S. in 2018. At the U.S. average rate of 1.13 fatalities per 100M vehicle miles traveled, we would have seen 37 Tesla-related fatalities on U.S. roads in 2018. But in reality there were 'only' 12. If Tesla is actively working to kill its customers, it appears to be less effective than other OEM's.

@ToolGuy Art, The Popular Mechanics estimate of "27 to 180 deaths as a result of rear-impact-related fuel tank fires" does not come anywhere close to the total number of Pinto-related fatalities - it's a small subset. (The analogous figure for "Verified Tesla Autopilot Deaths" would be 3 for the U.S. plus 1 for China.) I have not stated that Tesla should not be concerned with Autopilot-related crashes. The Pinto fuel tank design did not offer (to my knowledge) any active safety benefits. According to Tesla's published safety reports (see below), Teslas have a lower accident rate when Autopilot is engaged.

@ToolGuy I agree and am only comparing the fuel tank related deaths vs autopilot related deaths. Other accidents happen and Teslas on the whole seem to be very safe cars. Autopilot though seems to have been oversold with respect to capability and there is certainly the appearance of Tesla simply telling customers "You know, it can't really drive itself right, wink wink". GM's system seems to have safeguards built in that would help to prevent these abuses, though I doubt there are enough Super Cruise equipped vehicles on the road to draw a meaningful conclusion. The NHTSA has (or should at least) a keen interest in autonomous vehicle deaths, no matter how infrequent especially with respect to things like the car driving into the backs of stopped firetrucks. You can argue that it is really safer and what not, but perception is everything and there is a perception that these systems lend themselves to abuse and when they do fail, fail in a pretty catastrophic manner. The visibility makes government involvement more likely and Tesla should introduce safeguards for that fact alone...it is just good business to keep the government from acting here as well as getting to the bottom of what happened to improve the systems.

@ToolGuy Another Autopilot death today. They'll continue racking up until they recall this easily abused system and fix its deadly faults. https://www.bbc.com/news/technology-51645566?fbclid=IwAR3S6BZuo2bspItmHgXcJt4tC_4Iy_H5DKdMMjZ5g0kO6GGv4qAQTekBZQE

While those numbers paint a rosy picture of Tesla there are some factors that make those numbers seem better than they probably are. The average age of Teslas on the road is certainly very low, since they haven't had much volume until relatively recently. Meanwhile the average age of the US fleet as a whole is somewhere near 12 years. That means there are a lot of not so new cars out there that don't come even close to the occupant protection of the average 2 year old car. So yeah in a similar crash severity you are more likely to die in in a 2000 model year than a recent version of the same or similar model. Then you need to factor in the buyers/drivers of those cars. People who can afford new cars are are more likely to be more responsible and mature than someone rolling in a car they picked up at the BHPH lot. The same argument applies to the fire data. So yeah show me the Tesla numbers against cars of a similar age, not vs the average of the fleet in total.

Let's suppose Teslas are the world's safest cars. That's awesome, but does that, or is it enough to excuse its indiscriminate kills? Would you have a problem with "an update" that disables Autopilot, completely disengaging when drivers aren't paying attention to their own driving? If so, why?

@DenverMike I'd have no problem having a feature that disables autopilot when a driver is not paying attention. However, it shouldn't just shut down. There should be a process where it pulls the car over to the side of the road and parks. There are cases where the lack of attention could be due to a medical emergency. That's the safest way of dealing with it and it would provide additional incentive for drivers not to abuse it. Let's talk about another feature on cars that can cause deaths. High performance. While I don't have statistics in front of me to confirm it, I suspect cars over a certain power to weight ratio are more prone to kill indiscriminately. Should high-performance capability or features be banned because some people abuse them? Should a Mustang GT500 become a trim package with 150 hp and CVT because people are going to abuse anything else?

@mcs Despite what you claim, you're still defending Autopilot. Others, at least might find Autopilot exceptionally "handy" while eating in the car, texting and whatnot, even if they won't admit it. Yes it doesn't have to abruptly disengage. Gently pulling off the road would be a nice touch, if say there was no driver interaction/awareness. But any low-performance car, boring Toyota, etc, can be abused and mishandled, and obviously driven too fast and or aggressive for conditions/weather. The point is that's not the car itself doing the kills.

Scoutdude, Very valid point. Here is IIHS data on driver death rates by make and model. Be aware that there is a data lag, and a confidence interval - because statistics. https://www.iihs.org/ratings/driver-death-rates-by-make-and-model Some luxury models have *very* impressive (low) numbers. And there are some real surprises (ex. between brands) on mainstream models. Some of the differences are likely due to safety equipment, much of the difference would seem to be tied to driver behavior (but this is speculation on my part).