Autonomous Car Progress

[stopcrazypp]

Well, if you want to get super technical and start with the very first prototype testing in SF, sure you would be right. You want to go back to the first prototype testing in SF because it fits your agenda of claiming Waymo took longer to deploy in SF. But that is silly. That prototype in 2009 was completely different from the ride-hailing that they are deploying now now in SF. I think it makes more sense to start with when Waymo actually started testing the ride-hailing that they plan to deploy to see how long it takes to test and deploy a new service, not start with the first prototype.

And I am consistent with Tesla too. I would say Tesla started testing FSD Beta 1 year ago because that is when they released FSD Beta to testers. Obviously, Tesla started testing some earlier versions of FSD back in 2016. But I would not go back to AP1 and say Tesla took 8 years to test FSD Beta. That would be silly.

If you want to look at platform, since at least 2019 they have been testing with the ride-hailing focused Pacifica based vehicles in SF, the same ones they used in AZ:

Previous reports linked showed sightings in 2018, so likely even earlier than the video.

Following original point of conversation, SF in 2021 can't be used as an example of a model of Waymo expanding into a new city as you seem to be suggesting, given they have significant experience with this city already (not a blank slate like Chandler).

 

[diplomat33]

If you want to look at platform, since at least 2019 they have been testing with the ride-hailing focused Pacifica based vehicles in SF, the same ones they used in AZ:

But that is not the same platform that Waymo is using in the ride-hailing in SF. Waymo is using the I-Pace with the 5th Gen in SF.

That is why I start counting in 2021 because that is when Waymo scaled up testing of the platform that they are using for the ride-hailing in SF.

Previous reports linked showed sightings in 2018, so likely even earlier than the video.

Following original point of conversation, SF in 2021 can't be used as an example of a model of Waymo expanding into a new city as you seem to be suggesting, given they have significant experience with this city already (not a blank slate like Chandler).

Yes, Waymo did autonomous testing in SF way before 2021. But that testing was on older platforms before Waymo started work on deploy ride-hailing in SF. My point is that if you want to know how long it takes Waymo to deploy ride-hailing, you should start counting from when Waymo actually started deploying ride-hailing. 2021 is when Waymo started active testing for deploying ride-hailing in SF.

 

[EVNow]

But that is not the same platform that Waymo is using in the ride-hailing in SF. Waymo is using the I-Pace with the 5th Gen in SF.

So, we should just ignore all the work & testing done the first 4 gens ? I don't think so ....

Waymo/Google has been developing/testing for a decade now.

 

[diplomat33]

So, we should just ignore all the work & testing done the first 4 gens ? I don't think so ....

Waymo/Google has been developing/testing for a decade now.

No, not ignore. Obviously, the work done on the first 4 gens was key to get Waymo to where they are now. Stop trying to twist my words.

Here is the timeline:
2009: Waymo was founded.
2009 to 2018: Waymo developed autonomous driving.
2018: Waymo launched their first ride-hailing with safety drivers in Chandler.
2020: Waymo deployed driverless in Chandler.
2022: Waymo plans to deploy driverless in SF.

So it took Waymo 11 years to develop their autonomous driving from the Google prototype to where it was good enough for ride-hailing with safety drivers in Chandler. Then, it took Waymo 2 years to go from safety drivers to driverless in Chandler. It took another 2 years to deploy driverless in SF with 1 year of large scale testing with safety drivers in SF.

 

[S4WRXTTCS]

Yes, and your claims were...not accurate... as I laid out about that specific ruling in some detail.

It only happened in one country, not "countries" as you claimed.

It only happened in one regional (not national) court in that country.

It was about a court not understanding what the actual word autopilot means and instead imagining it "misleads" you into thinking it requires no driver- despite the fact the term has been used to describe an aircraft system for decades that still requires a pilot

And to my knowledge has resulted in no actual change in Teslas behavior in that country, as they still sell autopilot in Germany with the same wording as before the court case.

In fact- just as a quick sanity check on myself, I VPNed into Germany, went to Teslas website, changed my country to Germany there, and every word the court did not want Tesla to use is still being used 2 years later

Suggesting the verdict held no actual force of law (or was overturned by a higher court but I can't find a record of that)


So to sum up-- One regional court in one place misunderstood what the word meant and issued a bad ruling that didn't actually have any effect. Which is pretty far from your claim that multiple countries with consumer protection laws have used them to cause...anything at all... regarding AP or FSD.


View attachment 784571

Thanks for going through all that effort to see what the German site actually says.

But, the difference is clear as day.

It's not Full Self Driving, but Full potential for autonomous driving.

Now you gotta go back to work to use VPN plus an archive service to see what it said before the court ruling.

If Tesla advertised the package as Full Potential for Autonomous driving we wouldn't have any reasonable argument to expect to have L4 autonomous driving.

I guess I'll start referring to FSD as FPAD

 

[EVNow]

So it took Waymo 11 years to develop their autonomous driving from the Google prototype to where it was good enough for ride-hailing with safety drivers in Chandler. Then, it took Waymo 2 years to go from safety drivers to driverless in Chandler. It took another 2 years to deploy driverless in SF with 1 year of large scale testing with safety drivers in SF.

Doesn’t Waymo have reports of more than 2 years in CA ? So, they have been testing in CA for more than two years.

Just like AP/FSD is older than HW3.

 

[stopcrazypp]

But that is not the same platform that Waymo is using in the ride-hailing in SF. Waymo is using the I-Pace with the 5th Gen in SF.

It's the same platform they used in AZ for ride hailing which the I-Pace inherited a lot of things from. I think it's fair to leave out vehicles like the Firefly perhaps, but the Pacificas were designed for ride-hailing just like the I-Pace.

That is why I start counting in 2021 because that is when Waymo scaled up testing of the platform that they are using for the ride-hailing in SF.

Yes, Waymo did autonomous testing in SF way before 2021. But that testing was on older platforms before Waymo started work on deploy ride-hailing in SF. My point is that if you want to know how long it takes Waymo to deploy ride-hailing, you should start counting from when Waymo actually started deploying ride-hailing. 2021 is when Waymo started active testing for deploying ride-hailing in SF.

That's not a fair account either. Waymo has been testing extensively using the Pacificas since before then in SF. In 2020 they removed them from SF for a short while due to possible unrest:
Waymo pulls self-driving cars in San Francisco as cities nationwide brace for unrest

I have seen the Pacificas personally in SF during the pandemic.

The CA disengagement reports say for Waymo:
2014 53,428 miles (unknown model only Sept-December 2014 numbers)
2015 409,759 miles (unknown model)
2016 654,627 miles (unknown model)
2017 294,929 miles (unknown model, although note Pacificas were introduced this year)
2018 Pacificas 1,271,587 miles, I-paces 0 miles
2019 Pacificas 1,454,137 miles, I-paces 0 miles
2020 Pacificas 628,442 miles, I-paces 396.3 miles
2021 Pacificas 302,220 miles, I-paces 2,023,624 miles.

Leaving out 3+million miles the Pacificas tested on Bay Area and SF roads is misleading if you are trying to answer the question of how long it might take to deploy in a "new" city.

Here's the original question for context, person asking obviously already knew that Waymo had extensive experience in SF (so not surprising to him at all for it to take 6 months after ride hailing launch) and was asking about how long it would take for different cities which Waymo have never operated in:

Yes - even for Waymo anything more than 6 to 12 months would be dragging their foot for too long.

Infact when I just read the headlines about Waymo starting in SFO, I thought it was already driverless and paying customers.

What we really need to know is - how long does it take Waymo to expand to n+1 city. That will determine whether they can ever be successful.

 

[t3sl4drvr]

Waymo is absolutely stylish compared to the Apple autonomous car. Yikes. Could Apple have made the roof sensors any bigger? It looks like a king size mattress tied to the roof. LOL.

Makes sense to slab all the sensors one can on the car while gathering training data. This is to ensure they have the most accurate reference to train the target sensors with. The key question will be: what sensors will be included in the production car. (assuming they ever ship)

How much Apple will let the looks of the sensor suite dictate chosen sensors? Obviously they are not adding a "mattress" on the roof. They have a history of choosing form over function in some products. In the post-Jony era they have started to increasingly weight function. What will happen with the car?

Tesla has managed to choose their sensor suite extremely well when judged purely by how sensors look. Whether the chosen suite will ever be enough for L4 will be an extremely interesting to see.

 

[diplomat33]

Doesn’t Waymo have reports of more than 2 years in CA ? So, they have been testing in CA for more than two years.

Just like AP/FSD is older than HW3.

Yes, of course Waymo has been testing in CA for more than 2 years. Waymo spent a lot of years working on autonomous in CA driving BEFORE they deployed ride-hailing in Chandler or SF.

I guess the debate is: for the purpose of counting the time to deploy in n+1 cities, do we count all the years testing CA as part of the time to deploy in Chandler or as part of the time to deploy in SF. I count it as part of the time to deploy in Chandler since I am looking at things sequentially or chronologically. I am saying 11 years total to deploy in their first city (total time from founding to first driverless) and then 2 years to deploy in second city (time from first city to second city). You seem to be counting in the time to deploy in SF. But you can't double count those years. Either put them in the time to deploy in Chandler or in the time to deploy in SF, but not both. You calculate the time to deploy to n+1 cities from the last city deployed.

 

[diplomat33]

It's the same platform they used in AZ for ride hailing which the I-Pace inherited a lot of things from. I think it's fair to leave out vehicles like the Firefly perhaps, but the Pacificas were designed for ride-hailing just like the I-Pace.

That's not a fair account either. Waymo has been testing extensively using the Pacificas since before then in SF. In 2020 they removed them from SF for a short while due to possible unrest:
Waymo pulls self-driving cars in San Francisco as cities nationwide brace for unrest

I have seen the Pacificas personally in SF during the pandemic.

The CA disengagement reports say for Waymo:
2014 53,428 miles (unknown model only Sept-December 2014 numbers)
2015 409,759 miles (unknown model)
2016 654,627 miles (unknown model)
2017 294,929 miles (unknown model, although note Pacificas were introduced this year)
2018 Pacificas 1,271,587 miles, I-paces 0 miles
2019 Pacificas 1,454,137 miles, I-paces 0 miles
2020 Pacificas 628,442 miles, I-paces 396.3 miles
2021 Pacificas 302,220 miles, I-paces 2,023,624 miles.

Leaving out 3+million miles the Pacificas tested on Bay Area and SF roads is misleading if you are trying to answer the question of how long it might take to deploy in a "new" city.

Here's the original question for context, person asking obviously already knew that Waymo had extensive experience in SF (so not surprising to him at all for it to take 6 months after ride hailing launch) and was asking about how long it would take for different cities which Waymo have never operated in:

I am well aware of all the years Waymo tested autonomous Pacificas in CA and in SF. Yes, I am trying to answer the original question of the time it takes to deploy in n+1 cities.

I am looking at things chronologically. All the years of testing Pacificas in CA and in SF, before Chandler was driverless, is part of the "n=1" time. It took Waymo 11 years to deploy to 1 city (time from founding to deploying driverless in Chandler). It took Waymo 2 years to get to 2 cities, n=2 (time from driverless in Chandler to driverless in SF). You calculate the time to deploy to n+1 cities from the last city deployed.

So,

n=1, t = 11 years
n=2, t = 2 years

This makes sense since the n=1 time will include all the initial development time of building the autonomous driving from scratch. And we can expect the time for n=3, n=4 to probably go down as Waymo becomes more efficient in deploying robotaxis and gets closer to "solved FSD". However, there could be an uptick in the time to deploy if they deploy in a more difficult diving area or some external factor like a pandemic, natural disaster in the city or economic crash.

 

[Knightshade]

Thanks for going through all that effort to see what the German site actually says.

But, the difference is clear as day.

It's not Full Self Driving, but Full potential for autonomous driving.

Now you gotta go back to work to use VPN plus an archive service to see what it said before the court ruling.

If Tesla advertised the package as Full Potential for Autonomous driving we wouldn't have any reasonable argument to expect to have L4 autonomous driving.

I guess I'll start referring to FSD as FPAD

No I don't your own source already tells us what the court ruled on-

Tesla Germany is now banned from including “full potential for autonomous driving” and “Autopilot inclusive” in its advertising materials, Reuters reported.

Except, of course, they were never really "banned" from doing that- and the current website today uses that exact language for FSD as shown in the screen shot.

One regional court in one place made a wrong decision based on not even understanding what "autopilot" means, and that ended up not being enforced even 2 years later- that's all that happened.

I don't know enough about the German legal system to know if this is a case of the court didn't even have the authority to ban anything and just issued an unenforceable (and again, factually wrong) opinion on wording-- or if some higher body overturned them because they actually understood what the words Tesla was using mean. Not sure that really makes a difference either way though, since Tesla didn't end up having to do anything different and continues to sell the same product today using the same wording as then.

 

[diplomat33]

Saw something new I did not know about. In January 2022, the US Department of Transportation released the National Roadway Safety Strategy, which highlights how the Safe System approach can reduce the rate of transportation deaths in the United States:

Transportation Engineers (ITE), the Safe System approach is a shift to looking at traffic safety holistically, taking human error into account. It anticipates that human driver error, such as speeding, distracted driving, and drunk driving, will happen, and proactively puts duplicative safeguards in place that reduce critical injuries and fatalities from crashes that may occur.

Can Autonomous Vehicles Support the Safe System Approach to Traffic Safety? | Let's Talk Autonomous Driving

Waymo’s autonomous driving technology could support the Safe System approach in several ways, especially as it relates to the pillars of safe road users and safe speeds.

ltad.com

Here are the 6 principles of the "Safe System Approach":

A Safe System Approach incorporates the following principles:
1. Death and Serious Injuries are Unacceptable. While no crashes are desirable, the Safe System Approach prioritizes the elimination of crashes that result in death and serious injuries since no one should experience either when using the transportation system.
2. Humans Make Mistakes. People will inevitably make mistakes and decisions that can lead or contribute to crashes, but the transportation system can be designed and operated to accommodate certain types and levels of human mistakes, and avoid death and serious injuries when a crash occurs.
3. Humans Are Vulnerable. People have physical limits for tolerating crash forces before death or serious injury occurs; therefore, it is critical to design and operate a transportation system that is human-centric and accommodates physical human vulnerabilities.
4. Responsibility is Shared. All stakeholders – including government at all levels, industry, nonprofit/advocacy, researchers, and the public – are vital to preventing fatalities and serious injuries on our roadways.
5. Safety is Proactive. Proactive tools should be used to identify and address safety issues in the transportation system, rather than waiting for crashes to occur and reacting afterwards.
6. Redundancy is Crucial. Reducing risks requires that all parts of the transportation system be strengthened, so that if one part fails, the other parts still protect people.

Here are the 5 objectives:

Implementation of the NRSS will be arranged around five complementary objectives corresponding to the Safe System Approach elements:
• Safer People: Encourage safe, responsible behavior by people who use our roads and create conditions that prioritize their ability to reach their destination unharmed.
• Safer Roads: Design roadway environments to mitigate human mistakes and account for injury tolerances, to encourage safer behaviors, and to facilitate safe travel by the most vulnerable users.
• Safer Vehicles: Expand the availability of vehicle systems and features that help to prevent crashes and minimize the impact of crashes on both occupants and non-occupants.
• Safer Speeds: Promote safer speeds in all roadway environments through a combination of thoughtful, context-appropriate roadway design, targeted education and outreach campaigns, and enforcement.
• Post-Crash Care: Enhance the survivability of crashes through expedient access to emergency medical care, while creating a safe working environment for vital first responders and preventing secondary crashes through robust traffic incident management practices.

https://www.transportation.gov/sites/dot.gov/files/2022-02/USDOT-National-Roadway-Safety-Strategy.pdf

I think it is relevant because the "Safe System Approach" will likely play a role in AV safety regulations. It will likely affect AV deployment in general. The "Safe System Approach" will also likely influence how regulators look at Tesla's FSD when there is a crash. So it will affect Tesla's FSD approach.

 

[S4WRXTTCS]

I don't know enough about the German legal system to know if this is a case of the court didn't even have the authority to ban anything and just issued an unenforceable (and again, factually wrong) opinion on wording-- or if some higher body overturned them because they actually understood what the words Tesla was using mean. Not sure that really makes a difference either way though, since Tesla didn't end up having to do anything different and continues to sell the same product today using the same wording as then.

The problem with this entire discussion is the court ruling was about full potential for autonomous driving, and not FSD capability like I thought it was with my original post. For whatever reason I didn't catch that when I read the articles almost 2 years ago. Oops.

It makes everything I said in that post after rather comical as they were already using a phrase that I was arguing they should have used in the first place.

"Full self driving capability" is very different than "Full potential for autonomous driving"

To be technically accurate Germany (and maybe others) its FPAD, and in the US its FSD.

Now I'm still going to be stubborn by saying that if they used FPAD everywhere, and not FSD it would be a lot less misleading. To me potential isn't any kind of promise to achieve it.

Like Subway would be better off if they simply called their bread not really bread.

Subway bread is not bread, Irish court rules

Judge finds that sugar content of US chain’s sandwiches exceeds stipulated limit and they should thus be classified as confectionery

www.theguardian.com

 

[EVNow]

1. Death and Serious Injuries are Unacceptable.

Sounds like an empty political slogan to me.

 

[EVNow]

Yes, of course Waymo has been testing in CA for more than 2 years. Waymo spent a lot of years working on autonomous in CA driving BEFORE they deployed ride-hailing in Chandler or SF.

I guess the debate is: for the purpose of counting the time to deploy in n+1 cities, do we count all the years testing CA as part of the time to deploy in Chandler or as part of the time to deploy in SF. I count it as part of the time to deploy in Chandler since I am looking at things sequentially or chronologically. I am saying 11 years total to deploy in their first city (total time from founding to first driverless) and then 2 years to deploy in second city (time from first city to second city). You seem to be counting in the time to deploy in SF. But you can't double count those years. Either put them in the time to deploy in Chandler or in the time to deploy in SF, but not both. You calculate the time to deploy to n+1 cities from the last city deployed.

So,

n=1, t = 11 years
n=2, t = 2 years

No - that is obviously not correct. We can't "count" all the years Waymo did testing in SF as belonging to "AZ" - just to reduce the time one can attribute to SF. Only thing we can say truthfully is they have ten 13 years for 2 cities (actually small parts in 2 cities).

The answer to my question - what will it take for them to include (n+1)th city - can really only be answered by Waymo, if they have thought about it. Or we have to just wait.

SF isn't even saying they will start service in NY anytime soon - they are only talking about gathering info.

Anyway - I guess the first question is - when will Waymo cover the entire Bay Area - let alone a different metro.

 

[stopcrazypp]

I am well aware of all the years Waymo tested autonomous Pacificas in CA and in SF. Yes, I am trying to answer the original question of the time it takes to deploy in n+1 cities.

I am looking at things chronologically. All the years of testing Pacificas in CA and in SF, before Chandler was driverless, is part of the "n=1" time. It took Waymo 11 years to deploy to 1 city (time from founding to deploying driverless in Chandler). It took Waymo 2 years to get to 2 cities, n=2 (time from driverless in Chandler to driverless in SF). You calculate the time to deploy to n+1 cities from the last city deployed.

So,

n=1, t = 11 years
n=2, t = 2 years

This makes sense since the n=1 time will include all the initial development time of building the autonomous driving from scratch. And we can expect the time for n=3, n=4 to probably go down as Waymo becomes more efficient in deploying robotaxis and gets closer to "solved FSD". However, there could be an uptick in the time to deploy if they deploy in a more difficult diving area or some external factor like a pandemic, natural disaster in the city or economic crash.

But that's not how it works. You are counting things in "parallel". In that kind of counting, if waymo launches in two cities simultaneously (or close) then t=0 for one of the cities, but that kind of counting is worthless to determine how long it takes for Waymo to deploy in a city that it has absolutely zero experience in. Note I'm not saying that Waymo would necessarily take a long time, just that SF does not serve as a model given the experience Waymo has in it, even with a ride-hailing dedicated vehicle.

 

[bradtem]

I was going to point out this, but you beat me to it. Waymo has been operating in SF way longer than people are aware (it's not a "new" city for them), just not in the scale they are now. The Bay Area in general they have been in since before they went to AZ (I remember seeing their Firefly vehicle years ago).

For door-to-door L2 (AKA "feature complete" FSD) this is not out of the realm of possibility, in the same way people use AP in roads across the US that Teslas have never driven in before.

We don't trust our lives to AP on those new roads. We just ask it to keep the car in the lane and we are ready to catch it if it fails. That's very different from "Car, drive me through this complex city while I sleep even though you've never driven on these roads before."

 

[stopcrazypp]

We don't trust our lives to AP on those new roads. We just ask it to keep the car in the lane and we are ready to catch it if it fails. That's very different from "Car, drive me through this complex city while I sleep even though you've never driven on these roads before."

Door-to-door L2 doesn't allow you to sleep, you still have to monitor the car just like AP. Even L3 does not allow you to sleep, because you have to be ready to take over within seconds of the car requesting. Door-to-door L2 is pretty much the current end game for FSD Beta, at least according to Tesla's CA DMV filings. There are no plans yet for anything beyond that at the moment.

 

[daktari]

Even L3 does not allow you to sleep, because you have to be ready to take over within seconds of the car requesting.

This is wrong. The L3 car will stay in lane, then try to wake you, then safely stop and put hazard lights on and auto call emergency services if you fail to respond to a take over request.

One is not supposed to sleep in it anyway.

 

[stopcrazypp]

This is wrong. The L3 car will stay in lane, then try to wake you, then safely stop and put hazard lights on and auto call emergency services if you fail to respond to a take over request.

One is not supposed to sleep in it anyway.

Nope, you are incorrect, L3 by definition does not require the car to be able to stop safely on its own in all conditions, which is why you can't sleep in it. When it requires your attention, you must be ready to take over within seconds, which sleeping is not compatible with. I argued against a Jalopnik editor on this exact point before.
I'll just quote my comment here:

“For Level 3, this can mean that when the car is not capable of continuing independently and may require human assistance, it can safely pull over to the shoulder or outermost lane if the driver doesn’t respond. Level 3 systems still prefer a human to be monitoring them, though, and if they cannot get a human’s attention when requested, will safely stop.”

Sorry, I strongly disagree with this take on level 3 and your suggestion that people can safely sleep when using it. The description above describes a level 4 system, not a level 3. A L3 system is not required to safety stop/pull over, if it cannot get a human’s attention when requested (even if some systems claim they can in certain narrow conditions). You can ask Honda and I bet they will say you can’t sleep when using their L3 system.

In fact, I can save you the work, it says it right in Honda’s press release for their L3 system!

“At Level 3 of automation, the system monitors the driving environment surrounding the vehicle and takes over driving operations under certain conditions. When any of operable driving environment conditions become unsatisfactory, the system will issue a warning, and the driver must take over the driving immediately.”

Honda Global | November 11, 2020 Honda Receives Type Designation for Level 3 Automated Driving in Japan

Honda Global | Honda Receives Type Designation for Level 3 Automated Driving in Japan

global.honda

Here’s the definition of “LEVEL or CATEGORY 3 - CONDITIONAL DRIVING AUTOMATION” in SAE J3016 (5.4) (bold emphasis mine):

“The sustained and ODD-specific performance by an ADS of the entire DDT with the expectation that the DDT fallback-ready user is receptive to ADS-issued requests to intervene, as well as to DDT performance-relevant system failures in other vehicle systems, and will respond appropriately.”

SAE J3016 Table 2 describes the role of the “driver” in L3:

“Level 3 – Conditional Driving Automation

DDT fallback-ready user (while the ADS is engaged):

• Is receptive to a request to intervene and responds by performing DDT fallback in a timely manner

• Is receptive to DDT performance-relevant system failures in vehicle systems and, upon occurrence, performs DDT fallback in a timely manner

• Determines whether and how to achieve a minimal risk condition

• Becomes the driver upon requesting disengagement of the ADS”

Sleeping is very much not compatible with the above roles!

Examples given for “DDT performance-relevant system failures in vehicle” is a broken body or a suspension component (p 23), a failed a radar sensor that the system notices and request you intervene, or a broken tie rod (p10). If you are asleep, you won’t be able to notice those, but for L3 you are supposed to notice those and take over. All L3 does over L2 is give you a couple extra seconds to respond, not that it will be safe if you do not do so.

Again link to SAE J3016 for those who have not read it:

https://saemobilus.sae.org/download/?saetkn=7Ab97uLiFw&method=downloadDocument&contentType=pdf&prodCode=J3016_201806&cid=1000408663

The SAE Autonomy Levels Are Confusing But I Think I Have A Better Way