Hacker News

points by jedberg 1 day ago | 0 comments

> unsupervised autopilot

They are running a fleet of taxis around the Bay Area. It's kind of scary because they aren't that safe, given the lack of LIDAR. But they are doing it anyway.

[–] paxy 1 day ago

They all have someone behind the wheel. Tesla doesn't have a driverless permit in the Bay Area.

[–] jedberg 1 day ago

Yes, they have safety drivers for now, but they are still running them as an autonomous fleet.

And also, they are driving fully autonomous in Texas and Florida.

[–] paxy 1 day ago

> for now

So let me guess, fully unsupervised self driving is rolling out "next year" right?

[–] iknowstuff 1 day ago

They have more fully unsupervised robotaxis riding around Texas than Waymo.

https://www.tesla.com/robotaxi/safety

[–] katbyte 18 hours ago

That link doesn’t say how many, and it doesn’t differentiate between when there is a driver in the car or not

> Miles Driven Autonomously

[–] iknowstuff 12 hours ago

It does. None of them have a driver. Doesn’t count bay area.

[–] torginus 1 day ago

Photogrammetry + AI has developed to the degree, that there are 3D scanning apps on iPhone that don't even use the Lidar when available, so you can certainly make do without and get comparable results.

[–] jvwww 1 day ago

In the Bay Area they have human supervisors but in Austin they are autonomous.

Is there data to suggest they are unsafe in Austin - as in less safe than a human?

[–] timy2shoes 1 day ago

They're working hard to not release any of their safety data, unlike all the other autonomous car companies.

[–] iknowstuff 1 day ago

wrong

https://www.tesla.com/robotaxi/safety

[–] Slow_Hand 1 day ago

I’m not sure these self-reported numbers are trustworthy considering they report “0 robotaxi injuries” when a cursory Google search turns up this article claiming 249 Roboxtaxi accidents in Austin and at least two injuries that required hospital visits. Which is at least two more than “0 injuries”.

[–] iknowstuff 1 day ago

Those were Waymos…

[–] Slow_Hand 1 day ago

Yes. My mistake. It seems that the article is grouping all autonomous taxis under the title of "Cybertaxi", which is where I was confused.

[–] Geezus_42 1 day ago

That's what Elon wants you to think...

[–] iknowstuff 12 hours ago

A shining example of brainrot

[–] brightball 1 day ago

What makes you say they aren't safe? The numbers say they are about 4-6x safer than a car with a human driver.

[–] standeven 1 day ago

What numbers?

[–] iknowstuff 1 day ago

these https://www.tesla.com/robotaxi/safety

[–] tsimionescu 1 day ago

We know this data is trustworthy because Musk and his companies would never ever ever lie about anything important.

[–] iknowstuff 1 day ago

Ah you’re right it’s not like there’s any media scrutiny of robotaxis

[–] Geezus_42 1 day ago

Then share those instead of the same biased blog post.

[–] iknowstuff 1 day ago

https://www.tesla.com/robotaxi/safety

[–] Geezus_42 1 day ago

Biased source.

[–] jfoster 23 hours ago

What would an unbiased source be for this data?

[–] Geezus_42 18 hours ago

A blog post from the company definitely ain't it.

[–] jfoster 17 hours ago

Sorry, I meant the opposite. What would a good source be? Isn't Tesla the primary source of such data? Where else would you expect it to come from?

[–] Geezus_42 15 hours ago

Tesla can not be impartial about itself.

[–] jfoster 6 hours ago

So... where should the data come from?

[–] absoflutely 17 hours ago

A source that doesn't have an incentive to lie about the topic...

[–] jfoster 6 hours ago

Can you name it?

The implication seems to be that Tesla are lying about accidents. It seems like that would easily get caught out. Accidents tend to be things that happen in public.

[–] jfoster 23 hours ago

A fleet of hundreds of cars, right? Maybe even thousands including Austin? All with no LIDAR, and operating perhaps up to 18 hours a day?

So then where are all the horrible accidents if they're not safe?

[–] bushbaba 1 day ago

Why do you assume lidar is must for safety. So far the Tesla auto pilot record is better than human drivers

[–] katbyte 18 hours ago

Because less then ideal conditions for vision exist? Heavy snow/rain/fog, sun shining into cameras at dawn and dusk etc

It’s not hard to be better then an average human generally, the issue is liability as when a driverless taxi hits or hurts someone you can’t just go “but we are safer then human drivers!”

[–] arjie 1 day ago

I don’t see why it’s scary. Most cars on the road have a driver and no LIDAR. The cybercab is no different.

[–] jedberg 1 day ago

They intend to take the driver out, and they are already running them without a driver in Texas and Florida.

[–] small_model 1 day ago

Yep, very safely https://www.tesla.com/robotaxi/safety

[–] senordevnyc 1 day ago

Thank goodness we definitely know we can trust the data in that link /s

[–] iknowstuff 1 day ago

gov reports seem to agree so unless you have any data to deny this just take the L

[–] Geezus_42 1 day ago

Share those reports then.

[–] kalleboo 1 day ago

You mean the government that Elon Musk gutted with DOGE?

https://www.newsweek.com/doge-cuts-watchdogs-investigating-s...

> Elon Musk and his Department of Government Efficiency (DOGE) reportedly cut 30 National Highway Traffic Safety Administration (NHTSA) workers in February, disproportionately affecting regulators tasked with assessing self-driving vehicle risks, according to a Financial Times report.

[–] dozerly 1 day ago

For now, their goal is to have no driver and no lidar. That is what said the commenter is scared of.