As much as we have chafed at the limits and progress of FSD, it could be worse: Cruise agrees to cut fleet of San Francisco robotaxis in half after crashes | San Francisco | The Guardian General Motors’ Cruise autonomous vehicle unit has agreed to cut its fleet of San Francisco robotaxis in half as authorities investigate two recent crashes in the city. The state department of motor vehicles asked for the reduction after a Cruise vehicle without a human driver collided with an unspecified emergency vehicle on Thursday. “The DMV is investigating recent concerning incidents involving Cruise vehicles in San Francisco,” the DMV said on Saturday in a statement to the Associated Press. “Cruise has agreed to a 50% reduction and will have no more than 50 driverless vehicles in operation during the day and 150 driverless vehicles in operation at night.” The development comes just over a week after Californiaregulators allowed Cruise and the Google spinoff Waymo to operate autonomous robotaxis throughout San Francisco at all hours, despite safety worries spurred by recurring problems with unexpected stops and other erratic behavior. . . . Our often frustrating FSD experiences have helped to refine the system. But taking a step too far and fast, like 'cruise', could severely set everything back. There are FSD enemies, O'Dowd, Cummings, and some foolish politicians, who are ready and willing to kill it. So while we wait on V12, I have a backlog of maintenance to do to my 2019 Model 3 Std Rng Plus. Among my 'to do list' are a failed ultrasonic sensor, damaged under body panels, redesigned upper control arms, finding the mythical 'battery reserve,' and aerodynamic tuning. Bob Wilson
FSD in a recent acronym that is not in general use. Please spell it out at least ONCE in a post where it is used. Thanks. (What IS it ??)
Sorry, I'd copy and pasted my post from a Tesla forum. Full Self Driving (FSD) is currently at version 11.4.4 since June. Going more than two months without an update is unusual. Elon has reported that Version 12 will retire a huge amount of C+ code with more AI inference engines in today's weak areas including trip planning. Bob Wilson
A Cruise car tries to not collide with a fire truck, with front red lights on, heading towards it in the oncoming lane, and fails to do so. Authorities has GM reduce the number of operating cars in a city wide pilot program. A number that didn't exceed 300 cars on the roads of San Francisco. Teslas using Autopilot and FSD have collided with parked firetrucks, and even pedestrians. They seem to get into more crashes than all the other manufacturer's cars using an ADAS. That number of crashes did increase with the roll out of FSD. And Tesla continues to sell thousands of cars with FSD. Tesla 'Autopilot' crashes and fatalities surge, despite Musk's claims - The Washington Post
Perhaps because there are more Teslas by orders of magnitude with both AutoPilot and Full Self Driving. The NHTSA is collecting counts without converting them into rates either per vehicle or miles traveled. Meanwhile: Tesla Vehicle Safety Report | Tesla In October 2018, we began voluntarily releasing quarterly safety data in order to provide critical safety information about our vehicles to the public, and in July 2019 we began voluntarily releasing annually updated data about vehicle fires as well. Accident rates among all vehicles on the road can vary from quarter to quarter and can be affected by seasonality, like reduced daylight and inclement weather conditions. I am biased because I paid $2,000 in 2019 for Autopilot in my new Model 3. Two months later, it saved my wife, dogs, and me from an accident when I suffered a medical problem. It paid for itself that day. I now have over 4 years of AutoPilot and years of Full Self Driving. Although not perfect, their flaws are in the noise compared to how much safer my driving has been. So let me suggest this, share the accident rates in either per vehicle or miles traveled. Use the same scale Tesla is already using with both the numerator and denominator. ... The world waits in wonder. BTW, I recently bought a 2017 BMW i3-REx that lacks any Automated Driving System. It has dumb cruise control but it works great driven manually in traffic. The short wheel base, BMW suspension, and low battery biased weight, it is a pocket rocket great for around town. Comparing the two: Tesla Model 3 - excels on long distance trips and does OK around town. BMW i3-REx - excels around town and does OK on long trips. Bob Wilson
From the WP, "The uptick in crashes coincides with Tesla’s aggressive rollout of Full Self-Driving, which has expanded from about 12,000 users to nearly 400,000 in a little more than a year. Nearly two-thirds of all driver-assistance crashes that Tesla has reported to NHTSA occurred in the past year." That's a 33 times increase in cars with FSD, and a 3 times increase in crashes. From a 30% increase in car sales. Back to the OP These, FSD and Cruise, aren't comparable. GM's robotaxis are Level 5 automated cars being used in limited regions, with oversight from the local authorities to develop and certify the system. Despite the name and Musk's implications, FSD is just level 2, and as long as Tesla refuses the outside scrutiny for certification, that is all it every be. What has happened with Cruise here is just part of the process, and is not a severe set back for all.
Thanks for the link: Autopilot and Full Self-Driving (FSD) We have now released FSD Beta to nearly all customers in the US and Canada who bought FSD (approximately 400,000). This is an important milestone for our company. Every customer in the US and Canada can now access FSD Beta functionality upon purchase/subscription and start experiencing the evolution of AI- powered autonomy. That was about the time ("Q4 and FY 2022 Update") that the "safety score" went away. If you didn't have a high enough safety score, you could not get and enable FSD. Today, you can still 'strike out' but after 7-10 days, it is working again. But there is more from the same source: I'll see your "percentage" versus Tesla facts and data. I love percentages because if you have 1 car whose population increases by 33 times, that is 33 cars. At least then the 33 samples can be used for trustable conclusions ... of the 33 single cars. Bob Wilson ps. BTW, I bought FSD in October 2019 for $6,000. I have to keep my 2019 Model 3 because FSD is the most expensive part.
Musk announced recently that in (only) Q3, you will be able to transfer your FSD to another Telsa vehicle.
The front, bottom under the fender cover has scraped on curbs and barriers. I have a replacement but it has been too hot to install. The rear, bottom had a piece of road debris stuck in it. I could not pull it out so I cut it off and continued to my brother's home: This is what we pulled out. I have a replacement cover but again, too hot. Hind sight, I might have put one end of the split tube under a front tire and backed up to get it out. Bob Wilson
I linked this same article two months ago: WaPo: 17 fatalities, 736 crashes: The shocking toll of Tesla’s Autopilot Tesla’s driver-assistance system, known as Autopilot, has been involved in far more crashes than previously reported Also: "In a March presentation, Tesla claimed Full Self-Driving crashes at a rate at least five times lower than vehicles in normal driving, in a comparison of miles driven per collision. That claim, and Musk’s characterization of Autopilot as “unequivocally safer,” is impossible to test without access to the detailed data that Tesla possesses. Autopilot, largely a highway system, operates in a less complex environment than the range of situations experienced by a typical road user." Tesla didn't comment for that WaPo article, and it appears that this Tesla safety page has not been updated since. Or even since January. This sudden increase revealed this year, combined with last year's article about AP aborting control less than a second before crashes, leads some skeptics to wonder if this skews (shrinks) the claimed AP-associated crash rate, and raises more questions about the coverage and validity of the Tesla's claimed safety reports. Tesla should respond to these issues about its crash reporting. Have you seen any updated "facts and data" since then? Or any data at all since 2022Q4?
Ok, you win. You should not buy a Tesla 'death machine.' But you might consider taking a course in statistics. BTW, my Tesla is not for sale. I really like my Tesla death machine: 2023 Tesla Model 3 Top Safety Pick+ 2023 Tesla Model 3 Since I have Full Self Driving since Oct 2019, here is the NHTSA recall on Full Self Driving end of the report: https://static.nhtsa.gov/odi/rcl/2023/RCLRPT-23V085-3451.PDF So the problem has already been solved. My Tesla is at FSD V11.4.4 since June and I've already mapped out and handle the FSD problem areas. A level 2 system, I'm an "aware driver" who is allowed to correct FSD problems. Bob Wilson
I see your sleight of hand. Notice carefully: the 2023 Tesla Model 3 is just a 2022 Top Safety Pick+: 2022 TOP SAFETY PICKs It is conspicuously absent from the updated 2023 Top Safety Pick+ winners list: 2023 TOP SAFETY PICKs Fortunately for those who require this updated top rating, the 2023 Tesla Model Y, and 40 non-Tesla models, do have this 2023 rating.
Actually I used the IIHS search function. They have yet to report on the "2023" and my calendar says it is August 22, 2023. The IIHS has no problem reporting poorly performing cars. Absence could be they are still testing. Take your concern up with the IIHS as projecting something else ... Bob Wilson
Grasping at straws? The Model Y and plenty of other models were evaluated and made the 2023 "+" list without their current model year being crash tested. A half dozen models appear to have not made this list at the very start of the model year, but stepped up to it with production changes, after various dates ranging from October 2022 to May 2023. The Model 3 failed that same evaluation for the "+" rating, dropping out of the Top 40. At least it did retain the next tier score of Top Safety Pick. None of this addresses the suspicions that Tesla just might have been 'cooking the books' on its purported crash statistics, or why it seems that its primary public response since some of those suspicions bubbled up at WaPo has been to cease posting quarterly updates to that safety page. I very much doubt that an FSD software update will get the Model 3 stepped up to the IIHS "+" list, when this rating doesn't require AP or FSD in the first place. And Model Y made it on last year's software. Otherwise, I'll believe it when IIHS posts it. Several other maker's models stepped up to this list with production change dates months later than shown in this NHTSA posting
...... is the aborting of autopilot just one second before crashes implying that the AP code is deliberately doing this to avoid possibility of blaming crashes on autopilot? ...... or - might it (AP disengaging) happen because drivers that may not have been paying attention (a few seconds before likely crashes) determine an impending crash & therefore finally pay attention yanking the wheel / hitting the breaks etc .... which then disengages AP ? wouldn't the former (if true) be a type of fraud ? .
Tesla does say in their stats that it counts any crash in which Autopilot was deactivated within 5 seconds before impact. They also count any type of impacts so this means even when a vehicle is stop at a red light and it's rear ended.
That is one of the smells. Even if only the tin-foil-hat brigade smells it, it ought to be publicly addressed. Did I miss the response somewhere? "The NHTSA report also shows that "in the majority of incidents" among those 16 under close investigation, the Teslas activated their forward collision warnings and automated emergency braking systems, so it wasn't as though the drivers were given zero time to react, though it isn't mentioned how far in advance of impact those kicked on. In 11 of the crashes, none of the drivers took any action between two and five seconds before impact, indicating they, like Autopilot, didn't detect the impending collisions, either." [emphasis added] Linked here:
