Autonomous Car Progress

[diplomat33]

There are laws against fraud. There is no law against "misleading" the public. The entire science of advertising is how to mislead without breaking the law. And lawyers have made a science of how to seem to be saying something you are not. They can make a true statement appear to mean something entirely different. The ODD will be accurately described in five hundred pages of jargon so arcane that nobody can understand it, but which seems to the layperson to be describing our common-sense notion of "It will work anywhere a human would normally be willing to drive."

And even if they were caught lying and fined, when was the last time a big corporation had to pay fines more than about 1% of the profit they made with their lies? Never. The tobacco companies were proven to have knowingly lied about the deadly consequences of smoking, were proven to have known all along that millions of people would die of smoking, and lied about it, and they got fined an insignificant percentage of the profits they made from those lies, and the government even applied pressure on foreign countries to stop them from opposing smoking.

Don't count on the government to enforce any kind of openness on corporations.

I just don't think it will be as big of a problem than you think. It will be up to people to educate themselves of course. But we have to deal with the ODD for L2. L2 can have very different ODD's. We have to deal with the ODD of Autopilot, the ODD of Supercruise etc... Not sure why it would be significantly different for L4. I know you are not worried about Autopilot's ODD since you know that Autopilot works in Maui. But like I said, you'll know if L4 works in Maui or not. It won't be hidden.

 

[daniel]

I just don't think it will be as big of a problem than you think. It will be up to people to educate themselves of course. But we have to deal with the ODD for L2. L2 can have very different ODD's. We have to deal with the ODD of Autopilot, the ODD of Supercruise etc... Not sure why it would be significantly different for L4. I know you are not worried about Autopilot's ODD since you know that Autopilot works in Maui. But like I said, you'll know if L4 works in Maui or not. It won't be hidden.

But that's mainly because I know not to trust what the automakers say.

 

[avoigt]

I love this comment so much for various reasons. It is vacuous, lacking any remote inkling of logical thought or intelligence. You have been repeatedly corrected many times yet you keep on repeating it. This comment can literally be applied to any company and it would fit because it is a throwaway comment that does not have any insight. Keyword to remember is "intent", you were already told in the other thread.

Lets put it to the test with one tweak.

Cruise

Tesla

Zoox

Mobileeye


You know what all these companies have in common? They are all developing autonomous vehicle technology, none of them have any commercial autonomous taxi program but technologically speaking they are all engaged in developing technologies that will enable autonomous driving. They all also test L4 vehicles on public roads with safety drivers except for Tesla.

Which bring me to how inane your comment is.

Tesla


There is no public evidence that Tesla has a L4 test vehicle, they have only ever reported a L3 vehicle that drove 12 miles with zero disengagement. But we all know they are testing vehicles one way or another even when there has not been a single sighting of one. We assume they are testing in consumer vehicles when you use Autopilot but also they must also have vehicles with more advanced features that they test with as a technology company that builds things. Your comment states that they don't simply because they have a safety driver which for the hundredth time is not a prerequisite to something being L4. The design intent is what makes it so as long the vehicle can accomplish all tasks of driving safely within its operational design domain.

There is but 1 company that actually has and is testing an autonomous taxi service, and that would be waymo. We know this because they are required by law to report their data. From their data we can see that they give more than 6000 paid rides in their autonomous taxi service per month and of those roughly 300 is fully driverless without a backup driver per month.

Last week I interview the CEO and co-founder of Recogni, a hidden champion in autonomous driving that supposed to be in your above list of candidates to succeed and bring first vehicles on the road already in 2021.

Listen in to the first interview of R K Anand on my YT channel.

Exclusive: CEO & Co-Founder of Recogni, R K Anand, in His 1st Interview

 

[diplomat33]

Last week I interview the CEO and co-founder of Recogni, a hidden champion in autonomous driving that supposed to be in your above list of candidates to succeed and bring first vehicles on the road already in 2021.

Listen in to the first interview of R K Anand on my YT channel.

Exclusive: CEO & Co-Founder of Recogni, R K Anand, in His 1st Interview

From what I can tell, Recogni is a start-up, founded in 2017, that got $25M to develop "camera vision".
Recogni Emerges from Stealth Mode with $25 Million in Series A Funding to Deliver High-Performance, Low-Power AI Processing for Autonomous Vehicles

No offense but I am not sure how they are a "hidden champion". They are a start-up who just got money to develop "camera vision" that they claim will be faster and use less power. Nice but I can't find any evidence of how good their "camera vision" is. And they definitely don't have any autonomous driving. But you think they are a serious contender to bring autonomous vehicles to market next year? Unlikely. Just because a company says "give us money to build AI and camera vision" does not make them a contender for delivering autonomous driving. They are not in the same league as Waymo which already has L4 autonomous vehicles on public roads now.

 

[daniel]

So they just got funding to develop camera vision and before their work starts they're claiming that their system will be faster, cheaper, and better than existing systems? Reminds me of EEStor that, upon successful testing of a material they expected to be a suitable dielectric, announced that in six months (or did they say a year?) they would be mass-producing ultra-capacitors that would make lithium batteries obsolete. They didn't even have a prototype capacitor built. They just had a material they thought would work as the dielectric.

There are so many start-up in this field, all being touted as the sure winners. It's great that so many people are working on better batteries and better vision systems and better AI for autonomous cars. But I'd rather read "Here's another company that's going to work on this. Welcome to the field and good luck," rather than read "This new company that just formed last month is going to have a self-driving car with a 500-mile-battery in half a year."

 

[Bitdepth]

Last week I interview the CEO and co-founder of Recogni, a hidden champion in autonomous driving that supposed to be in your above list of candidates to succeed and bring first vehicles on the road already in 2021.

Listen in to the first interview of R K Anand on my YT channel.

Exclusive: CEO & Co-Founder of Recogni, R K Anand, in His 1st Interview

You have a very interesting channel. I perused your channel and you have some genuinely interesting interviews. I did a little digging and the co founder of Recogni Ashwini Choudhary is a serial entrepreneur who likes to create companies for acquisition. He creates companies that look good on paper so bigger companies buy them out. Just a cursory look at previous companies a lot of their upper staff have been involved with seems like silicon valley funding startups. I'm not saying they don't or won't have a product to sell but a hidden champion in autonomous driving they most definitely are not.

 

[Bitdepth]

Tesla driver spots a driverless waymo
Imgur
https://i.imgur.com/ySOnXJx.mp4

 

[diplomat33]

Tesla driver spots a driverless waymo
Imgur
https://i.imgur.com/ySOnXJx.mp4

Thanks. It's great to see.

Waymo is like "Hi Tesla! I'm a real robotaxi!"

 

[Tikib2020]

"Hi waymo! I am a real product, not a financial black hole!"

 

[mikes_fsd]

Tesla driver spots a driverless waymo
Imgur
https://i.imgur.com/ySOnXJx.mp4

What a cluster-!@#$% of a setup.
See all those additional sensors, and especially all those additional moving parts.
That is what is called additional points of failure and a maintenance nightmare.

no-gracias.

 

[thewishmaster]

What a cluster-!@#$% of a setup.
See all those additional sensors, and especially all those additional moving parts.
That is what is called additional points of failure and a maintenance nightmare.

no-gracias.

That's why they're not selling these to consumers

 

[diplomat33]

What a cluster-!@#$% of a setup.
See all those additional sensors, and especially all those additional moving parts.
That is what is called additional points of failure and a maintenance nightmare.

no-gracias.

It's not a cluster-f@$@. Waymo did not just throw a bunch of additional sensors on the car for no reason. The sensors are configured that way on purpose to allow the car to perceive everything around it in all conditions. It's what makes it possible for the car to drive itself so safely without a human driver.

 

[mikes_fsd]

It's not a cluster-f@$@. Waymo did not just throw a bunch of additional sensors on the car for no reason.

Keep telling yourself that.

They threw all those sensors on for one main reason...
It being: distract everyone else from realizing that they do not have a viable/scalable/economical solution to FSD.

 

[diplomat33]

Keep telling yourself that.

They threw all those sensors on for one main reason...
It being: distract everyone else from realizing that they do not have a viable/scalable/economical solution to FSD.

Wrong. You clearly don't understand autonomous driving.

 

[mikes_fsd]

You clearly don't understand autonomous driving.

What is clear, is that I fully disagree with your opinions on autonomous driving.

 

[timewasted]

What is clear, is that I fully disagree with your opinions on autonomous driving.

You appear to disagree with reality. Waymo operates vehicles that literally have no one driving them, and you still say that they don’t have self driving cars.

Do you know what you call all those sensors that they have on their vehicles? Progress! Autonomous driving is an unsolved problem regardless of what some cokehead in Fremont likes to proclaim. The proper approach to solving it is iterative. If there’s a sensor available that can help you make progress, you use it. Only once you have a solution to the problem can you realistically backtrack and figure out what pieces of the system you can remove.

Mere decades ago the device that would be required to write this comment would take up floors in a building, but today it fits comfortably in the palm of my hand. At some point in the future we will see a similar transformation for autonomous vehicles, but that won’t happen until we have a system that is proven to actually work. THAT is the problem that Waymo is working on, and actually making progress with.

 

[mikes_fsd]

You appear to disagree with reality.

Boy, I don't want to know what your reality is really like.
Let me know when you're approved to ride a Wyamo car, I signed up on their app, and I am "not approved at this time".

Waymo is a really expensive PHD thesis project for Google, not a business plan that is being brought to life.

 

[Soda Popinski]

I think people are arguing past each other. Mspisars is not arguing Waymo hasn't demonstrated FSD / autonomous driving / SAE Level 4. He's arguing they don't have a fully functional robotaxi service, as it's not open to the public yet, and is ugly.

 

[daniel]

Developing FSD is a very long-term project. One step along the way is an extremely expensive prototype that actually works, in a limited geographical area and limited conditions. Nobody here thinks that Waymo has reached the goal of a self-driving consumer car. What they have done is reached a significant milestone: A super-expe car capable of driving in city traffic without a driver, correctly responding to all the confusing signals and all the pedestrians, cyclists, road hazards, construction zones, and pedestrians.

Still ahead is refining it and bringing down the cost and expanding the areas it can operate in. These are huge hurdles. But what they've done so far is impressive.

And I don't give a rat's behind what the car looks like. Heck, my Zap Xebra was one of the ugliest cars on the road and I loved it.

 

[Daniel in SD]

Developing FSD is a very long-term project. One step along the way is an extremely expensive prototype that actually works, in a limited geographical area and limited conditions. Nobody here thinks that Waymo has reached the goal of a self-driving consumer car. What they have done is reached a significant milestone: A super-expe car capable of driving in city traffic without a driver, correctly responding to all the confusing signals and all the pedestrians, cyclists, road hazards, construction zones, and pedestrians.

Still ahead is refining it and bringing down the cost and expanding the areas it can operate in. These are huge hurdles. But what they've done so far is impressive.

And I don't give a rat's behind what the car looks like. Heck, my Zap Xebra was one of the ugliest cars on the road and I loved it.

This is what always puzzles me about the notion that Tesla's strategy for FSD is obviously the best. I can't think of a technology where such an approach worked. Tesla made the Roadster first, expensive and bespoke, before making the still expensive Model S and eventually the mass market Model 3. The first cellular phones cost $10k and were enormous.
If you look at the history of AI, smart people have been making overly optimistic predictions for over half a century. We'll see if this time is different.
History of artificial intelligence - Wikipedia