We now know TESLA has world leading robotaxi safety. 250,000 Robotaxi Miles driven with 1 incident. Austin reveals the safety and Tesla revealed the miles.
Putting the pieces together. We have the up to date public safety data. Tesla revealed the miles of robotaxi in Austin and in the SF Bay Area. We put this together with the known start dates for multiple data points for the ramp and a nearly perfect safety record.
Tesla revealed the Austin robotaxi miles driven as 250,000 and most of those were Q3. Q2 only had 7K miles.
Tesla revealed SF Bay Area robotaxi miles as over 1 million. There was only a tentative ramp in August. Most of the miles had to be on September and first half of October.
We now know there a lot miles of clean and safe robotaxi driving. This public data for Waymo and Tesla robotaxi can now be compared. We also know the exponential ramp is 15X the miles from Q2 to Q3. Miles are doubling every 2-4 weeks.
City expansion and safety driver removal and vehicle expansions are all driven by continuing the safe operation. The cleaner the safety record then the faster the expansion.
The cities that need to be added and the monthly pace and actions for cities added and FSD version improvements are now clear.
The Q2 release of Tesla Cybercab required massive Tesla confidence. This will be thousands of vehicles that are for the sole purpose of commercial unsupervised robotaxi.
Brian Wang is a Futurist Thought Leader and a popular Science blogger with 1 million readers per month. His blog Nextbigfuture.com is ranked #1 Science News Blog. It covers many disruptive technology and trends including Space, Robotics, Artificial Intelligence, Medicine, Anti-aging Biotechnology, and Nanotechnology.
Known for identifying cutting edge technologies, he is currently a Co-Founder of a startup and fundraiser for high potential early-stage companies. He is the Head of Research for Allocations for deep technology investments and an Angel Investor at Space Angels.
A frequent speaker at corporations, he has been a TEDx speaker, a Singularity University speaker and guest at numerous interviews for radio and podcasts. He is open to public speaking and advising engagements.
Hm, magically somehow my post disappeared. Here it is again.
Folks, my very honest suggestion is- find a statistician. The comparison you have made does is very superficial is not interpretable.
Here is why:
1. Comparing apples to oranges. Waymo is fully autonomous, Tesla is not. You need to add critical engagements.
2. Data you show does not match NHTSA reports of at least 3 accidents in Austin prior to end of September. Why?
3. Multiple confounding factors. How are areas you compare matching? Is Waymo driving in areas that are more dangerous than Tesla (or vice versa). You need to calculate normalization factor for that or account in another way (add in a linear regression model)
4. Same of time of day
5. Type of accident. At fault, not at fault, technical issue (Waymo had a wheel malfunction)
6. Estimates are not acceptable for this type of comparison as Scott pointed out
Overall you would need to do a lot more work to be convincing.
Both versions 13 & 14 are estimated, not actual, miles until Disengagement (DE). Further, in the chart: https://app.powerbi.com/view?amp;pageName=ReportSectiona513b9781c5673d6f4fb&r=eyJrIjoiZTlkMWY1NTMtYTM1Mi00YmZmLWE4ZjQtZmE0M2EyMjgzMDNjIiwidCI6ImMxM2M0MmQ1LTlhNTAtNDY3YS05Yjc3LWI1MjkyYzgxZjM1NSIsImMiOjF9&pageName=ReportSectione34408b53060e1deb61f ALL version 14 interations COMBINED total just 3,006 miles as of this writing, which is too little draw any conclusions about the 100% @DriversNoCritDE (No critical disengagement). And this obscures the far less %DriversNoDE, which we’re told to believe would not have escalated into critical disengagement if not engaged in by humans: averaging ~62%, but actually lower is averaged on a weighted basis since the most recent version of 14x has only 8 miles in the chart.
Rapid version upgrades obscure the true reliability and safety profile of the Robotaxi (not even the Robocab, which may have entirely different characteristics, and in any case, no interventionist passenger/driver either).
As Foyle pointed out, there is essentially no improvement in a year.
Liability clarification – in court – is just beginning to shake out, with a few recent cases going against Tesla for multi-million dollar settlements. Even if a tiny percentage <0.001% of miles driven, this would become an unsustainable business model if scaled up to 100s of 1,000s of miles per month eventually. And non-geofencing driving, which is the only useful measure in major taxi-dependent cities like NYC, will increase the odds of some kind of accident exponentially. In NYC, we decreased the speed limit to 25mph from 30mph, on most streets, which lessened fatality, but not enough to prevent a sizable and influiential lobby trying to reduce it further to 20mph. At 25mph, most ebikes, certainly class 3 ebikes, can operate on roads, and there's a lobby to make them totally illegal (they are already illegal in the state, outside NYC). Pedestrians and bike riders alike have seen increased accidents, even fatalities. Vision Zero is a failure. For now, the liability falls on the driver, but if cars are autonomous? Who pays? If there are multiple vehicles involved, no-fault insurance is the only way forward, but the state or the insurance companies would have to be convinced autonomous vehicles are safer than driver-vehicles, in order to cover costs. Is there any evidence this is happening?
Eh, DRivesNoDE, not DriversNoDE. A human mistake… Meh.
I drive a hardware 4 model Y, about 60 miles/day in a giant metropolitan area. Software version 13.2.9.
I fully trust the software. The only interventions I’ve done the last few months are highway lane preferences and pothole avoidance. None of them were ‘critical’ or because I was afraid for my safety.
I can’t wait for 14.3. If it’s unsupervised, I’ll be comfortable relinquishing control.
But robotaxi is all under continuous supervision, taking care of any near misses and mistakes, Waymo is (as far as I am aware) unsupervised. So this really is meaningless.
Open source tracking of critical disengagements on the 4 versions of FSD14.x released so far show that it is has made no significant improvement in disengagements over 13.x or 12.x of a year ago, yet. It is going to take at least one hardware update and a huge amount of AI development to eliminate the 999 out of 1000 current disengagements needed to get safe enough for unsupervised driving.
https://app.powerbi.com/view?amp;pageName=ReportSectiona513b9781c5673d6f4fb&r=eyJrIjoiZTlkMWY1NTMtYTM1Mi00YmZmLWE4ZjQtZmE0M2EyMjgzMDNjIiwidCI6ImMxM2M0MmQ1LTlhNTAtNDY3YS05Yjc3LWI1MjkyYzgxZjM1NSIsImMiOjF9&pageName=ReportSectione34408b53060e1deb61f
Dont let facts get in the way of a good story