[clouddrover]

> As a general rule, if Tesla isn't doing it already I suspect it's either not economically feasible, difficult to manufacture, or otherwise has drawbacks they've judged signifigant.

That's not a very good rule. There a lots of things Tesla doesn't do.

- Tesla doesn't do battery swaps, but NIO does. NIO's done it 10 million times: https://www.nio.com/blog/nio-users-china-have-completed-10-m...

- Tesla doesn't do 350 kW chargers, but Ionity (among others) does. Current EV truck drivers appreciate it: https://www.youtube.com/watch?v=9UAttTG03WA

- Tesla's charging network is slowly rolling out charging to all EV brands in Europe, but Tesla still doesn't charge all brands in North America. Electrify America (among others) does.

- Tesla doesn't offer EV scooters, but Gogoro (among others) does: https://www.gogoro.com/gogoro-network/

- Tesla doesn't offer EV vans, but other manufacturers do.

- Tesla doesn't offer EV city cars, but other manufacturers do.

- Tesla doesn't do a heads-up display, but other manufacturers do.

And so on.

[gridder]

- Tesla doesn’t offer LiDAR sensors, even if it’s clearly the best technology to use for the autopilot.

[jeffreygoesto]

Radar is working hard to get towards higher resolution and classification quality. From what I know, it is the closest competitor because it is active sensing and complements camera properties even better than LIDAR. In adverse conditions LIDAR and Cameras have similar failure modes, which is not good.

I do agree that cameras an ML on the images seems to be stuck at the next plateau of scene understanding performance. An impressive one, admitted, but still not sufficient. For real emergency reactions with highway speeds you need to assess situations about a quarter mile away basically.

[gzer0]

Not even close. Psuedo-LiDAR approach is the way to go and everyone knows it. Waymo and others using LiDAR placed the wrong bet and now have nothing to show for it.

[chell830]

> Waymo and others using LiDAR placed the wrong bet and now have nothing to show for it.

Waymo launched their commercial robo-taxi service in 2017. It’s still running today. Open to the public [1].

Meanwhile, Tesla is phasing out their FSD program. Beta-testers, 98 safety score, $15K cost, whatever, current Tesla owners are pissed off.

FSD is a sinking ship, just like the Tesla/Elon brand. They used to be cool.

[1] https://www.azcentral.com/story/money/business/tech/2017/04/...

[drcross]

You obviously haven't seen the FSD upgrade that was released yesterday.

[hahaxdxd123]

Haven't you people been saying this for 2 years now?

[jjulius]

The update that is only to v10.69, instead of v11 like Musk had suggested would roll out now? The same upgrade that's only rolling out to ~1k FSD-enabled customers (which is only 1% of FSD Beta customers)?

[gridder]

I hope you’re not biased since you work for Tesla’s autopilot BU. Anyway I’m no expert and I’m open to learn something. With pseudo-LiDAR do you mean photogrammetry? Following your input I may say the best approach could be then to integrate both image and LiDAR technologies to create the most trustable 3d scan. Since LiDAR measures space and distance, it can correct the other photography sensors which could be prone to some errors. I understand the cost side of scale manufacturing but you cannot cut cost in this case, the risk of fatal injuries it’s too big. Also I feel Elon decided ‘dogmatically’ to avoid LiDARs and that’s why this is the only road you are considering…

[ohgodplsno]

You work on autopilot at Tesla. You are _the_ demonstration that Tesla's camera based approach is insufficient, has awful deficiencies and is a dangerous product. I wouldn't be saying much if I were you.

[dmitriid]

Except, you know, in parts of the world that are not sunny California. Your camera will mean nothing in the Nordics/Canada with snow/slush/dirt on the road, and the sun shining straight at you.

But sure, do tell me how you're going to ignore the laws of physics and see a road sign, a kid, or a deer when your camera is blind.

[sschueller]

Waymo hasn't killed anyone yet. I think that is probably the most important thing to show.

[rcMgD2BwE72F]

Waymos drive so slow, any Tesla with the same "driver" would kill anyone else.

[arcticbull]

Well, except not running over mannequins of kids. [1] You know, the simple stuff.

[1] https://www.youtube.com/watch?v=6aMWERzT7SQ

[drcross]

This has been completely debunked numerous times in the last week. It's a total smear campaign by a competitor.

[hahaxdxd123]

oh https://twitter.com/TaylorOgan/status/1557766646026084359

[thedrbrian]

>This has been completely debunked numerous times in the last week

show us

[awestroke]

Because it is prohibitively expensive, and you get more data with multiple cameras than with one single lidar

[clouddrover]

It didn't prohibit Mercedes:

https://www.caranddriver.com/reviews/a39966189/2022-mercedes...

https://electrek.co/2022/07/22/mercedes-drive-pilot-review/

[awestroke]

Well yeah, if you double the price of the car you can afford to slap lidar on it. You'll sell fewer cars, however

[clouddrover]

So what's the Model S's and Model X's excuse? They've got the Mercedes EQS price tag but not the equipment.

[awestroke]

Tesla want to offer FSD on their whole lineup, not just in their high end models

[__m]

Not at night

[dmitriid]

> you get more data with multiple cameras than with one single lidar

Until you're at night in the snow/dirt.

[hoseja]

Yeah, lidar famously deals well with inclement weather and low visibility conditions. Those infrared lasers just punch right through the fog and snowflakes.

Lmao.

[dmitriid]

Ah you decided to be smart and sarcastic, and inadvertently showed just how limited the tech in these "autonomous" cars is.

And that you can't just handwave these problems away.

[hoseja]

Is this a sarcastic joke? Surely you can't mean that.

[neither_color]

On July 4, 2022, at the NIO power exchange station at Shanghai Hongqiao VEG Micro Creative Park, an ES8 marked the 10,000,000th power swap in NIO’s history.

Impressive that they were able to do this in July of '22 after just being in one of the world's most severe lockdown for months in a row. There were almost no non-essential vehicles on the road in Shanghai for at least 60 days

https://edition.cnn.com/2022/07/13/china/china-covid-shangha...

[rsynnott]

I mean, presumably the Shanghai facility is just the one that happened to do the 10 millionth change; China's a big country. Also, obviously, there was a time before the lockdown, when they presumably racked up some battery changes.

For comparison, if Amazon or someone claimed they'd just shipped their billionth package, to someone in Luxembourg, "there aren't many people in Luxembourg so this sounds wrong" would be a weird thing to say.

[Smoosh]

> And so on

I would really like V2H or V2G but no sign of it from Tesla so far.

[clouddrover]

Or even just V2L. Honda has it with the Honda e, Ford has it with the F-150 Lightning, and Hyundai has it with their E-GMP platform (currently Kia EV6, Genesis GV60, and Hyundai Ioniq 5).

[dzhiurgis]

IMO that should be illegal to sell. Such a missed opportunity.

[judge2020]

Technically all of those are covered under "otherwise has drawbacks they've judged significant", but I think they meant "in the realm of EV performance": the Plaid is still the fastest car you can buy without dropping a half million on a SF90 or even more on a limited-run supercar.

[zuhsetaqi]

It's the fastest in acceleration. Bring it to the Nürburgring and it's by far not the fastest car. Even the Porsche Taycan Turbo S is more than 2 seconds faster[0]. Driving quality or how good a car is is not defined by how fast a car can accelerate.

[0] https://www.nuerburgring.de/info/nuerburgring/records

[arethuza]

I was surprised by just how much faster the fastest ICE cars are than the fastest electric cars - about 50s faster?

[mhh__]

Electric cars are really really heavy, and people looking for a trackday monster are going to buy a 911.

[zuhsetaqi]

Yes, probably because electric cars have the heavy battery on board

[clouddrover]

> Technically all of those are covered under "otherwise has drawbacks they've judged significant"

Pray tell, what are these significant drawbacks?

> the Plaid is still the fastest car you can buy without dropping a half million

Nope: https://www.theverge.com/2022/8/19/23312956/lucid-air-sapphi...

[googlryas]

According to that article, that car isn't available yet and is only going to made in extremely limited numbers.

[clouddrover]

And? How many people do you think buy high performance cars? The answer is: not many.

[googlryas]

Someone said

> the Plaid is still the fastest car you can buy without dropping a half million

And you said:

> Nope: https://www.theverge.com/2022/8/19/23312956/lucid-air-sapphi...

Except you can't actually buy the car you linked to. That's a pretty big miss in a conversation about cars you can buy. In fact, I would say one of the two defining features of "cars you can buy" is the fact that you can buy them.

[clouddrover]

Of course you can. Reservations for the Sapphire open on Tuesday. Deliveries start next year.

Don't feel too bad that Tesla doesn't lead here. Tesla doesn't lead in real world range tests either:

https://insideevs.com/reviews/443791/ev-range-test-results/

Tesla makes good ads though. I like the bit with the parrot:

https://www.youtube.com/watch?v=NfMtONBK8dY

[threeseed]

You can reserve the Lucid now and have it available in early 2023.

I would suspect that it isn't too different from the Plaid.

[melony]

Ridiculous. Just walk around the Bay area and see for yourself.

[judge2020]

> Lucid Air Sapphire will be offered as a limited-production model, with deliveries planned in the US and Canada next year. The price is $249,000 USD and $325,000 CAD.

https://www.lucidmotors.com/stories/introducing-sapphire-pin...

[clouddrover]

Yes. That's not half a million now, is it?