Self-Driving Cars

So they’re selling a taxi service?

Because as long as the taxi service is functioning as advertised that doesn’t seem like a Beta product. More that they’re using one product (the taxi service) to develop another product (the self-driving).

In my understanding, Waymo is not currently charging for those rides. There are quite a few regulatory hurdles to go through to offer taxi service, but I don't know where they are in that process.

If they're not charging, that's even further from making people pay for a beta product.

I was referring to Tesla when I was talking about "paying for beta".

Wait… If they’re giving something away then how is it them selling something?

Like, we do agree that the concept of selling requires an exchange of money/goods, right?

What?

I said “ I’m still amazed that you can even sell a beta product for something like a car,” and you responded “ Talk to Waymo about that.”.

I know you like Tesla, but you shouldn’t be making up stuff about other self-driving developers. Or did you just skip over the word “sell” in my post?

Edited by Silasw on Oct 15th 2021 at 5:32:08 PM

I think we miscommunicated. I was referring to the regulatory environment of operating vehicles on public roads using developmental software, regardless of whether the occupants are paying for the privilege. There is a difference: Waymo is offering a commercial taxi service, while Tesla is selling to private customers, but there are still self-driving cars on the roads.

Edited by Fighteer on Oct 15th 2021 at 12:38:29 PM

The difference is that one of these presents a bigger loophole for selling unfinished products that bypass regulatory scrutiny than the other, which was what Silas said?

Again, these concerns seem based on ideology and speculation rather than any factual analysis of what's going on. "I don't trust what Tesla is doing" is one thing, but that's not the same thing as "Tesla is not safe".

Edited by Fighteer on Oct 15th 2021 at 12:41:11 PM

Aren’t all/most vehicles eventually tested on public roads before being put up for public sale?

You’re right that there’s still a gap there, with members of the public traveling in vehicles utilising software that hasn’t been through regulatory approval.

What’s going on is that regulatory gaps are being carved out. History shows us how well that tends to go.

Your ideological support for Tesla doesn’t change the fact that large businesses finding ways to avoid their products being subject to regulatory oversight always ends badly eventually, even if the first company or two are good/smart enough to not sell something defective.

Edited by Silasw on Oct 15th 2021 at 5:45:13 PM

Now you're just putting words in people's mouths.

"It's surprising this is allowed" does not automatically mean someone's said "this particular instance is unsafe".

Edited by RainehDaze on Oct 15th 2021 at 5:45:44 PM

If we are staking out positions, the reasons I support Tesla are because:

• It is single-handedly driving the adoption of electric vehicles at a far faster rate than would have occurred otherwise.
• It is attempting to solve self-driving using a vision-based, machine-learning approach that, if successful, will scale far better than more traditional approaches, such as lidar with high-definition mapping.

Further, if Tesla is successful at both of those goals, its self-driving solution will be immediately deployable to nearly all of its vehicles in operation, whereas anyone else would have to solve it first and then produce cars that are suitably equipped.

In short, I believe that it is uniquely positioned to solve both vehicle emissions and traffic deaths in a way that no other company can. For this reason, it is faced with extraordinary and virulent opposition from parties whose status quo stands to be disrupted and this opposition comes at it from all sides. It is crucial to recognize such biases in media coverage.

While there are certain to be safety issues on the path to fully autonomous driving, we cannot be paralyzed by these fears and, while regulation is important, we cannot allow it to sabotage development of these products, because real lives are at stake with every hour we delay.

Edited by Fighteer on Oct 15th 2021 at 1:09:39 PM

I don’t see why one needs to be either specifically supporting or opposing a company as a whole. It’s a big company and it does multiple things, I judge each actions individually.

I support Tesla’s development of electric cars, I support the development of self-driving technology, I oppose what I view as shady marketing and testing practices, I support cars being made more easy to upgrade via giving them software that can be patched via over-the-air updates, I oppose the constant pissing contests with regulators in a country that is well know for under-regulation.

I could go on, but I think my point is made.

Here's another interesting article that came to my attention via Twitter. From the IoT Times: 5GAA argues C-V2X technology as the way to secure transport at the ITS World Congress. The article is summarized and discussed in this Slashdot article.

The gist is that a representative of a 5G consortium is arguing that deployment of high-speed wireless data in cities will be essential before autonomous vehicles can properly operate there. Obviously they have an agenda: to get 5G deployed, but I've seen this idea before and I'm not sure where it's coming from.

From what I gather, there is an assumption that autonomy will require high-definition mapping (in conjunction with lidar sensors). Since no vehicle can carry enough onboard memory to store an entire region's worth of maps, never mind refresh them, it will require a high-speed wireless connection to transmit the necessary data, hence 5G.

However, those assumptions fail to take into account machine vision/neural net solutions like Tesla FSD, which do not rely on external data other than GPS and low-definition Google maps for route planning. While there may be various motivations for this omission, I really think that they're shooting themselves in the foot if vision wins the autonomy race.

Me, I'd also wonder about the reliability of a wireless system. If the onboard memory fails, that's one car (partially, if the self-driving car can still be steered the old-fashioned way) out of commission; if the wireless system fails, that's an entire city out of commission?

Yeah, always online self-driving cars sound like both a pain in the arse and a safety hazard.

Plus data storage is pretty good these days, can you really not get an entire city onto a car?

German language article but some discussion of what could happen if hackers gain access to self-driving cars. Wireless systems in particular could become an inviting target as you can hit many vehicles at once. Presumably hackers would do this to extort individual drivers/cities, and the risk is probably increasing as hackers and internet criminals get ever more sophisticated.

While I cannot read the article myself, I should note that this seems to be more of the "5G" thing. Everyone seems to be assuming that autonomy will require a constant wireless connection to feed real-time data to vehicles while ignoring that one of the leading solutions does not require this.

Well, yeah, but historically humans are really bad at pre-emptively recognizing and fixing problems. So it's quite possible that some localities adopt the unsafe solution and mayhem ensues. Seen it a million times from aviation design to politics to economic policy.

Sure, and it's also true that centralized autonomy is easier to develop from a programming standpoint. It is likely, for example, that Waymo will declare that it has arrived at a functional SAE Level 5 solution prior to Tesla doing so. However, getting Waymo-equipped cars deployed in large numbers to cities is going to be very slow and very expensive. They certainly won't be in private hands; they'll be used almost exclusively as robo-taxis where the cost can be rapidly amortized. For this same reason, we won't see the price of robo-taxi service go down very quickly if at all.

Meanwhile, if/when Tesla FSD becomes Level 5, there will be millions of cars on the road already equipped with it that can simply turn it on and go. It'll seem like it happened by magic and everyone will have vapors about how they were "caught by surprise".

Edited by Fighteer on Oct 19th 2021 at 7:33:50 AM

Given how many times Tesla announced level 5 automation, and their deployment strategy, it would probably no more catch the public by surprise then Apple maps finally working properly.

That would only apply if Tesla had a significant media presence. As it is, the lack of any sort of advertising means that an awful lot of people will be caught by surprise. If you are invested (intellectually or financially), you understand that there cannot be a straightforward timeline because it's literally a new technology being invented on the fly.

There's also no guarantee that it will actually work and satisfy regulatory requirements for SAE Level 5. So it's a gamble, but one that I feel is worth taking. Worrying about exactly when it'll happen or "broken promises" is missing the point.

Oh come on, Tesla has a huge media presence, sure it’s a social media/making news/viral marketing presence rather than an advertisement based presence, but it’s one of the best knows car brands out there.

Not having a advertisement purchasing budget is not the same as not having a media presence, Games Workshop has the same PR strategy.

The big announcement on level 5 won’t be technological anyway, it will be a legal announcement. That may be what some developers are counting on, that they can afford a gap between technological completion and their cars being on the roads because it’s going to take the law a chunk of time to catch up, so they can use that time to physically build the cars.

Hell, even a map based system doesn’t require that, it’s just a way to save on cost/weight. The idea of cars that require a constant network connection is madness, from a safety, security and user comfort perspective.

Edited by Silasw on Oct 19th 2021 at 8:17:40 PM

And an infrastructure one. It's just an utterly wasteful use of bandwidth.

Oh god, imagine the signal blackouts if a self-driving car convention comes to town…

"Chaos ensues as self-driving cars lose signal and are unable to reach their destination. Their drivers, unable to access online mapping or call ahead, were similarly flummoxed."