Waymo

[Bitdepth]

It's the rate of advancement, the absolutely most critical point, that you left out.

Nothing was left out. The rate of advancement is the same among everyone developing this technology. In the beginning it is always fast then it peaks and flattens for a long time because of the match of nines. Some are just in a more advanced state than others.

Tesla first released their infamous demo in Nov 2016 they've been working on

Tesla Self-Driving Demonstration | Tesla Canada

www.tesla.com

In 2016 this is what Waymo was doing

Waypoint - The official Waymo blog: A record month in miles

waymo.com

Lately our cars have been getting a lot more practice. In August alone, our fleet of 58 vehicles traveled a record monthly total of 170,000 miles; of those, 126,000 miles were driven autonomously (i.e. the car was fully in control).

Waypoint - The official Waymo blog: Accelerating the pace of learning

waymo.com

Our report shows a marked improvement in our fully self-driving technology. Since 2015, our rate of safety-related disengages has fallen from 0.8 disengages per thousand miles to 0.2 per thousand miles in 2016

Can your Tesla drive 1000 miles without an intervention or you touching the steering today, 7 years later from when they started? It took Waymo 7 years to go from no self-driving, to doing 0.2 disengagements per 1000 miles.

6 months ago, FSDb drove like a 6 yo. Now, arguably a 16 yo after a few lesions. It's the rate of advancement that is crucial, and as the training of the neural nets advances, that will accelerate at a far greater pace than other solutions can keep up with.

Do I need to remind you that 6-year-olds can't drive? I don't know what 6-year-olds you know that drive you are comparing FSDb to. But the rate of advancement is the same and, in many ways, behind where Waymo was 7 years after they started the Google Self-Driving Car Project.

 

[bkp_duke]

Appropriate article relevant to this discussion, today, regarding scaling and the issue with HD maps. Work fine . . . till something happens that isn't predicted and the system then can't cope.

Cruise Robotaxi Gets Stuck In The Middle Of An Intersection, Elon Musk Replies

This isn’t the first time the outspoken CEO replied to random Twitter posts, but it’s the first time we can think of that he got involved with a video showing a stuck Cruise Chevy Bolt EV.

insideevs.com

Yes, this is Cruise, but same principle applies to Waymo. Both are using the same approach.

 

[EVNow]

Appropriate article relevant to this discussion, today, regarding scaling and the issue with HD maps. Work fine . . . till something happens that isn't predicted and the system then can't cope.

Cruise Robotaxi Gets Stuck In The Middle Of An Intersection, Elon Musk Replies

This isn’t the first time the outspoken CEO replied to random Twitter posts, but it’s the first time we can think of that he got involved with a video showing a stuck Cruise Chevy Bolt EV.

insideevs.com

Yes, this is Cruise, but same principle applies to Waymo. Both are using the same approach.

CA should cancel these robotaxis until they can figure out how to be not stuck or have safety drivers. These cars are bringing a real bad name to AVs.

 

[spacecoin]

Appropriate article relevant to this discussion, today, regarding scaling and the issue with HD maps. Work fine . . . till something happens that isn't predicted and the system then can't cope.

Yes, this is Cruise, but same principle applies to Waymo. Both are using the same approach.

I find it hilarious and absurd that Musk implies that Tesla would handle this situation successfully, and that Tesla FSD is somehow better than the Waymo or Cruise. Or that Tesla has some approach that is vastly better or different for vision than these guys. It's at best the same tech, but in a budget package folks!

The only thing that is brittle is the bubble he's pumping up regarding FSDb. No redundancy, no multimodality equals no autonomy and people like @EVNow which himself (in the very forum) stated doesn't believe in camera-only in many years, keeps bashing on the competition.

If you have an engineering background, you should be able to understand how hard it is to remove the driver, and how far Tesla is from doing so. I takes about a year for each order of magnitude of improvement in reliability (looking at all the efforts that has come before) and Tesla isn't even at the 99% yet...

What will you be saying in three years when Tesla still can't drive autonomously in the LVCC loop? In the mean time Tesla will probably be launching "Really actual super-smart summon" but not ship it - and still shifts blame to the customer. Perhaps the auto-wipes work by then though? It took Tesla over 3 years to fix the darn auto high beam...

 

[hooty]

Don’t worry we have 9 months of rain here and rarely does the FSD not work.

‘Rarely’ ain’t gonna appease the authorities (if ever Tesla applies for a license)

 

[hooty]

6 months ago, FSDb drove like a 6 yo. Now, arguably a 16 yo after a few lesions.

No, it’s still a 6 year old, with psychosis

 

[EVNow]

‘Rarely’ ain’t gonna appease the authorities (if ever Tesla applies for a license)

Authorities are not exactly pleased with Waymo - the city would ban Waymo if they could tomorrow.

Anyway - what you are saying is wrong. When conditions deteriorate to outside ODD, the car has to safely park on the side - MRC. That's all that is required and obviously not what robotaxis are doing now as the sit blocking traffic all the time.

 

[bkp_duke]

‘Rarely’ ain’t gonna appease the authorities (if ever Tesla applies for a license)

Clearly you haven't kept up on news. SF is trying to pass an ordinance to ban both Waymo and Cruise from the city streets.

 

[SO16]

I find it hilarious and absurd that Musk implies that Tesla would handle this situation successfully, and that Tesla FSD is somehow better than the Waymo or Cruise. Or that Tesla has some approach that is vastly better or different for vision than these guys. It's at best the same tech, but in a budget package folks!

The only thing that is brittle is the bubble he's pumping up regarding FSDb. No redundancy, no multimodality equals no autonomy and people like @EVNow which himself (in the very forum) stated doesn't believe in camera-only in many years, keeps bashing on the competition.

If you have an engineering background, you should be able to understand how hard it is to remove the driver, and how far Tesla is from doing so. I takes about a year for each order of magnitude of improvement in reliability (looking at all the efforts that has come before) and Tesla isn't even at the 99% yet...

What will you be saying in three years when Tesla still can't drive autonomously in the LVCC loop? In the mean time Tesla will probably be launching "Really actual super-smart summon" but not ship it - and still shifts blame to the customer. Perhaps the auto-wipes work by then though? It took Tesla over 3 years to fix the darn auto high beam...

Yes Elon “implies” that FSD could handle it today, which of course it cannot.

The difference is that there’s a PATH that FSD “could” handle it “someday” where as the HD map method never will.

Now wether or not FSD gets there, that remains to be seen. Elon’s FSD time predictions have been poor.

 

[bkp_duke]

Yes Elon “implies” that FSD could handle it today, which of course it cannot.

The difference is that there’s a PATH that FSD “could” handle it “someday” where as the HD map method never will.

Now wether or not FSD gets there, that remains to be seen.

Last I heard, there was also an "approval" problem with the City of LV about using FSD in the loop. They had not signed off on it, so Tesla was not allowed to even try it yet.

 

[spacecoin]

Yes Elon “implies” that FSD could handle it today, which of course it cannot.

The difference is that there’s a PATH that FSD “could” handle it “someday” where as the HD map method never will.

Now wether or not FSD gets there, that remains to be seen. Elon’s FSD time predictions have been poor.

What you and many others seem to be confused about that that Waymo doesn't require the HD-map to be correct. They would not have a business if that was the case and there are plenty of examples where Waymo operates around roadwork these days. The map just one input to improve the driving when it is correct. It's nice to be able to know what's _likely_ around the next corner. Both maps and Lidar are a safety and reliability play. And if driverless vision only becomes viable in the future, they can remove the need for them if they reach the reliability and drive quality needed without them.

Do you think Tesla has better performance than Waymo on vision only? Not a chance.

What do you think Tesla does in CV that Waymo didn't do three years ago?

 

[spacecoin]

Last I heard, there was also an "approval" problem with the City of LV about using FSD in the loop. They had not signed off on it, so Tesla was not allowed to even try it yet.

So if it wasn't for the authorities Tesla would be (supervised) autonomous in the LVCC loop? No chance, tbh. They can't even get smart summon to work reliably. Again, ingress and egress are not trivial problems.

 

[bkp_duke]

So if it wasn't for the authorities Tesla would be (supervised) autonomous in the LVCC loop? No chance, tbh. They can't even get smart summon to work reliably. Again, ingress and egress are not trivial problems.

Hyperbole and conjecture on your part. Not facts.

 

[spacecoin]

Hyperbole and conjecture on your part. Not facts.

Whatever man. If Tesla is driverless autonomous in any small ODD (like the LVCC including pickup and dropoff) in two years I'll give you a 2:1 on any amount you'd like to wager. They should have gone for a train in the tunnels.

 

[bkp_duke]

Whatever man. If Tesla is driverless autonomous in any small ODD (like the LVCC including pickup and dropoff) in two years I'll give you a 2:1 on any amount you'd like to wager. They should have gone for a train in the tunnels.

Last guy here that bet with me, lost (badly). Then he squelched on the bet. Since then, mods have told me no more wagers. Your Euro are safe.

 

[spacecoin]

Last guy here that bet with me, lost (badly). Then he squelched on the bet. Since then, mods have told me no more wagers. Your Euro are safe.

Let's make it a bit playful then. If I am right you need to change your signature for a year to whatever I say, and vise versa.

 

[SO16]

What you and many others seem to be confused about that that Waymo doesn't require the HD-map to be correct. They would not have a business if that was the case and there are plenty of examples where Waymo operates around roadwork these days. The map just one input to improve the driving when it is correct. It's nice to be able to know what's _likely_ around the next corner. Both maps and Lidar are a safety and reliability play. And if driverless vision only becomes viable in the future, they can remove the need for them if they reach the reliability and drive quality needed without them.

Do you think Tesla has better performance than Waymo on vision only? Not a chance.

What do you think Tesla does in CV that Waymo didn't do three years ago?

1. Waymo doesn’t operate until they have HD maps for a given area. So Waymo feels they are needed. How much they are needed, I’d only be guessing. But that is a requirement. Yes there are examples of Waymo going around construction but Waymo says they can make quick updates to maps. Maybe they did that for those scenarios where it worked. Question is, does that scale?

2. Yes I think Tesla’s vision only is better than Waymo’s vision only due to video processing. But I think Waymo’s vision + LiDAR has better accuracy. Question is, is it needed? I think not if the AI can do better routing. My disengagements have been due to “bad calls” in routing, not because of hardware visibility limitations.

 

[bkp_duke]

Let's make it a bit playful then. If I am right you need to change your signature for a year to whatever I say, and vise versa.

Worthless to me. Your postcount is what I do in like a week, or less. My sig is worth 50X yours.

 

[spacecoin]

1. Waymo doesn’t operate until they have HD maps for a given area. So Waymo feels they are needed. How much they are needed, I’d only be guessing. But that is a requirement. Yes there are examples of Waymo going around construction but Waymo says they can make quick updates to maps. Maybe they did that for those scenarios where it worked. Question is, does that scale?

Waymo uses ML and fleet learning to update the maps these days. Waymo doesn't operate in unmapped areas because they wouldn't probably have the safety margins at this point to remove the driver, which is 100% of their business case. Limiting the ODD is also about being able to validate it

Of course it scales. 95% of the ride hailing market is in dense urban centers incl airport. .

2. Yes I think Tesla’s vision only is better than Waymo’s vision only due to video processing. But I think Waymo’s vision + LiDAR has better accuracy. Question is, is it needed? I think not if the AI can do better routing. My disengagements have been due to “bad calls” in routing, not because of hardware visibility limitations.

Due to vision processing? Can you elaborate what you mean by that? Given that Tesla is at less than 95% completed rides without a critical disengagement and a DE every 10-15 miles, I'd say it's needed for the foreseeable future for driverless deployment and/or L3+.

 

[JB47394]

SF is trying to pass an ordinance to ban both Waymo and Cruise from the city streets.

Do you have a link for this? The city is clearly not happy with being the testing ground for the two companies, and they're critical of expansions, but I can't find anything about an attempt at a ban. At least, not one that has caught the attention of the web.