Fatal Accident: How Is BMW Misleading Us with Autonomous Driving? | Hacker News Reader

Fatal Accident: How Is BMW Misleading Us with Autonomous Driving?

thelastdriverlicenseholder.com

12 pointsZhenya28 comments

I don't understand the argument the author is trying to make. The vehicle in question is not an autonomous vehicle, but rather a data collection vehicle that is collecting data about road conditions, environment, etc. in order to make use of that data as part of BMW's autonomous vehicle project. All of this is explained at the URL that is literally written on the vehicle: http://www.bmw.com/autonomousdriving

The author acknowledges having read that information, yet for some reason concludes that BWM is deliberately trying to mislead the public. The reason for that leap of logic is not obvious to me.

Outrage-oriented journalism? (I haven't read and probably won't read TFA).

Bloggers aren't journalists.

Looks like his not-a-journalist just tries to sell his book.

Submarine article funded by Tesla to add smoke around their issues with autopilot perhaps?

I honestly have no idea, but I'm becoming less surprised as time goes on.

Slightly off topic: I am the owner of a Tesla, and until recently a 2021 model 5 series BMW. The 5 series had 'keep your lane' and 'follow' functionality, and I've been absolutely flabbergasted that this functionality made it on production vehicles. It would give up in the middle of basic turns in the road in such a way that it tried to turn you straight into the divider. It was comically bad, except that it wasn't funny.

The Tesla (not a fanboy btw) is miles ahead in this regard.

> The 5 series had 'keep your lane' and 'follow' functionality, and I've been absolutely flabbergasted that this functionality made it on production vehicles. It would give up in the middle of basic turns in the road in such a way that it tried to turn you straight into the divider. It was comically bad, except that it wasn't funny.

Could that be deliberate to keep the driver attentive? I have a Honda with similar functionality, and it will just completely disengage if the driver hasn't adjusted the steering in 15-20 seconds. It also has automatic emergency breaking that deliberately will not prevent a collision, it will only slow you down. The effect of their decisions is that long distance driving is easier, but it's impossible to even try to let the car drive itself.

AFAICT it was totally random. I would keep my hands on the wheel at all time but even then it felt very surprising. After a couple of times I never engaged it again, except for couple of times to show others how ridiculously bad it all was.

Suggestion: the planet is dying. While I applaud your move to an EV, understand that your want for a new vehicle every couple years has an absolutely massive carbon footprint.

To anyone reading this comment: please buy used and understand that most problems with cars can be fixed economically by someone with a bit of knowledge.

> understand that your want for a new vehicle every couple years

you're making a lot of assumptions here. In general I'd agree with you.

Article comes across as super paranoid, and really attributes the worst possible conspiracy-theory level motives to a seemingly innocuous situation.

Calling bs on this article. It is normal that cars like that one can be used by the developers even off-hours. Typically you have to write a short test report after the weekend and ingest the recorded drives. That extends the spread of tested scenarios.

The car is not equipped for higher automation levels, which require a much bigger sensor set. You'd be able to see them. It has a bog standard assistance set.

The friggin' stickers are for GDPR reasons and they just designed a single sticker that is used all over the test fleet. The claim "automated" is the same hybris all over the industry, managers trying to compete on claims. It does say nothimg about the capabilities of BMW in the field.

100% wanted to state that. The stickers are also not as misleading. After all, the data collected will be used for that purpose, training or tests for autonomous driving.

I interpreted sticker as an add basically, nothing more.

I'd also like to second this and I'd like to add regarding the following paragraph:

"It is normal that cars like that one can be used by the developers even off-hours. Typically you have to write a short test report after the weekend and ingest the recorded drives. That extends the spread of tested scenarios."

That is absolutely true and BMW camouflaged cars with stickers, like the ones described in the article, are a very, very common sight in and around Munich. I'd say in the northern parts of the city and the Autobahn leading northward you are bound to see one if you just wait a short while. This also means that naturally they are involved in accidents and occasionally even in fatal ones.

Despite that these cars are common and there are a ton of people driving them, every single person moving one of these cars has had very thorough training. BMW has a whole bouquet of internal drivers licenses with increasing difficulty. It is normal that even people working in software have the basic levels of BMW driving licenses if there is the chance that the they have to use a "test vehicle" (which again there are different classes of).

It is sort of weird that http://www.bmw.com/autonomousdriving redirects (well, it does now) to https://www.bmw.com/en/footer/data-processing-automated-vehi...

Is it weird? Pretty sure all of BMW's first level path urls are redirects. They subdivide their actual urls by language.

There seems to be confusion as to whether the car was autonomous or not. The URL that is affixed to the car says "autonomous", but it redirects to one that says "data processing automated". Those two apparently English phrases mean different things.

What I'm curious to know is how BMW's L2 compares to Tesla's Autopilot (L2). I'm surprised there aren't more youtube videos of showing if it can maintain lanes properly. Did the BMW autosteer into the oncoming traffic or was it driver error?

BMW's L2 is pretty good 99% of of the time. I can't remember when I had to intervene the last time. The only issues it still has are with those low boulders/walls that divert a lane at construction sites or so, especially when coupled with confusing temporary lane markings that criss-cross the original ones.

That said, if I understand correctly, the accident the (pretty nonsensical) article refers to wasn't caused by autosteering. A human was steering the car at the time.

Just like with dieselgate this will be a closely guarded secret. Most likely the eu will create some very friendly regulation, and once it becomes an issue they will blame on east europeans somehow - there’s always a way in the eu for that.

> Just like with dieselgate this will be a closely guarded secret.

So ... no secret at all and widely published and discussed?

Once investigated by the americans, almost buried by merkel, no one seriously punished in europe, millions of polluting vehicles still on the roads and somehow blaming east europeans for buying said highly polluting vehicles as second hand goods - despite governments in those countries wanting to prevent such imports, but pushed back by the eu.

So yes, the same. No secret, published and discussed while nothing really done.

>Just like with dieselgate this will be a closely guarded secret

Dieselgate has a 25,000 word Wikipedia article with almost 500 citations. What secrets are being guarded?

Pretty well guarded by certain carmakers until the us revealed it. 25.000 words and no real consequences. No damages paid to european buyers, no serious fines across the eu for damage done to the environment, no consequences for people killed as a result of years of increased pollution, and vw is still in business.

No damages paid? Please get your facts straight: https://www.cleanenergywire.org/factsheets/dieselgate-timeli...

Reply on news.ycombinator.com