Autonomous Car Progress

[diplomat33]

So, you think to expand to a new city Waymo has to figure out if there are any edge cases in that city, train CNN for the new edge cases - and change their planning code - apart from generating HDMaps and manually annotating the path the car should take on every road / parking lot ?

As Waymo gains more and more experience, and solves more and more edge cases, their autonomous driving will work reliably in more and more places. It will take less and less time to expand to the next city.

Whenever Tesla gets to L4/L5 robotaxi in US - within a year or two they will expand to EU, China etc where they sell cars now.

Unlikely. Even if Tesla solves L4/L5 in the US, there are regulatory hurdles in the EU, not to mention different roads, different traffic rules in both the EU and China. There will be entirely new edge cases. To get L4/L5 working in the EU and China will require more retraining of the NN. It won't happen in one year.

What are the HD maps various autonomous ventures are now using ? I'm guessing they are very detailed not that 40 cm accuracy ones.

I don't know for sure but the accuracy is probably around 1-10 cm. So they are more accurate than Toyota's 40 cm HD map.

 

[daniel]

Cost to operate a robotaxi is under 5 cents a mile. Cost for a driver is somewhat under a dollar a mile. So instead of little cheaper it will be an order of magnitude cheaper. Agree this will take 10 years.

If you just look at operating cost you're ignoring the capital cost of buying the cars. I'll bet those Waymo cars cost about a million dollars each.

And as Doggydogworld comments above, 5¢/mile is a fantasy even excluding capitals cost and depreciation. The Uber:robotaxi comparison is the capital cost of the car vs. the labor cost of the driver. But in addition, there's still a lot of labor cost with a robotaxi as Doggydogworld also explains above.

And lastly, we still don't even know if a reliable robotaxi can be built capable of operating over more than a few very small ideal areas. Outside of places like Chandler, it will have to be able to handle snow, bad roads, and bad drivers. And there will still be riders who need help from a person. Robotaxis will be just one more option in the overall transportation system, which will include private cars, taxis, ride-shares, busses, etc.

 

[Terminator857]

Fuel and tires alone cost that much. Then add cleaning, inspections, maintenance and repair, monitoring, customer service, etc. And that excludes the biggest cost - depreciation. Add it all up then double the result to account for deadheading.

$25K tesla car depreciation over a million miles is 2.5 cents a mile. Deadheading has nothing to do with cost to operate, that is just less income. OEM tires cost about $50. Lets say they last 40K miles. That makes cost of tires half a cent a mile. My tires last longer so I'll assume a soft riding robotaxi will get even more miles on a set of tires.

 

[diplomat33]

I'll bet those Waymo cars cost about a million dollars each.

No. That is not true. Krafcik said the cars with all the hardware cost about the same as a well equipped Mercedes S-class. So maybe $150k at most. Definitely not $1M. Come on now!

 

[Terminator857]

What are the HD maps various autonomous ventures are now using ? I'm guessing they are very detailed not that 40 cm accuracy ones.

Waymo describes their 3d maps as centimeter accurate.

 

[ThomasD]

What about maintenance/cleaning costs. Paying people to take care of the cars. Insurance and registration fees and refueling costs gas or electric. Remote monitoring Leasing/purchasing costs of buildings for remote monitoring and storage/maintenance facilities

 

[ThomasD]

Would things like snow banks affect the the centimeter accurate 3d maps

 

[Terminator857]

... Leasing/purchasing costs of buildings for remote monitoring and storage/maintenance facilities

Since robotaxi operates 24x7 there is no need for storage facility. On a more serious note, there are plenty of unused parking spaces at 3 am. If shopping centers have electric car chargers then that could be a win.

 

[rxlawdude]

Would things like snow banks affect the the centimeter accurate 3d maps

Seeing how Waymo operates exclusively in snow free zones and where weather is relatively moderate, that's a great question.

 

[diplomat33]

Would things like snow banks affect the the centimeter accurate 3d maps

Are you asking if snow banks makes it harder to create HD maps or are you asking if it the snow banks messes with the car's ability to read the HD maps?

 

[EVNow]

Cost to operate a robotaxi is under 5 cents a mile. Cost for a driver is somewhat under a dollar a mile. So instead of little cheaper it will be an order of magnitude cheaper. Agree this will take 10 years.

Here is the actual analysis done by ARK in 2019. No it is not 5 c / mile. The range is between 20 and 47 cents.

 

[ThomasD]

Both Snow banks are ever changing. They can get smaller/larger in the space of a few hours. Day time melt then refreeze. You get more snow. The snow bank gets larger. Maybe a lot larger depending on snow fall amount and roads getting plowed making a small snow bank massive in just a few hours

 

[EVNow]

As Waymo gains more and more experience, and solves more and more edge cases, their autonomous driving will work reliably in more and more places. It will take less and less time to expand to the next city.

So, from Chandler to Sanfrancisco they took 3 years (?). How long before they add the next city ? How long before they add all top 100 US cities ?

Both Snow banks are ever changing. They can get smaller/larger in the space of a few hours. Day time melt then refreeze. You get more snow. The snow bank gets larger. Maybe a lot larger depending on snow fall amount and roads getting plowed making a small snow bank massive in just a few hours
View attachment 717509

Thats an edge case they didn't see in Chandler. They won't in SF.

 

[diplomat33]

So, from Chandler to Sanfrancisco they took 3 years (?). How long before they add the next city ? How long they before they add all top 10 US cities ?

They took 3 years to go from Chandler to SF because the FSD was still very new. It won't take that long to add the 3rd, 4th, 5th city etc... The time between adding new cities will get shorter each time. But how long it takes to add all top 10 US cities is anybody's guess.

 

[EVNow]

They took 3 years to go from Chandler to SF because the FSD was still very new. It won't take that long to add the 3rd, 4th, 5th city etc... The time between adding new cities will get shorter each time. But how long it takes to add all top 10 US cities is anybody's guess.

You didn't directly answer my question. When will Waymo add the next city ?

If they take 2 yrs 11 months - that would be less than 3 years. But not by much. But, if they add the next city in 3 months - thats a big improvement.

Its not just the idea that it will take decreasing number of years between cities that is important - but also the pace of decrease.

Let me put it in other words. Waymo has NOT shown that they can scale their technology to be a viable robotaxi operator. Full stop.

 

[diplomat33]

You didn't directly answer my question. When will Waymo add the next city ?

Nobody knows. If I had to guess, I would say probably 2 years maybe. I think Waymo will probably go full public in SF in about a year and then they will probably want to see how it goes before adding a 3rd city. But that is just my educated guess.

The fact is that there is nothing really stopping Waymo from adding a third city now (with safety drivers) if they really wanted to. It just depends on when Waymo's co-CEOs decide to do it.

Dolgov did tweet about Waymo learning a lot from driving autonomously in LA. So we know Waymo is doing autonomous driving in LA. Maybe Waymo is already looking to LA as the next city after SF? But that is just speculation on my part.

You all want to know when Waymo will expand to a 3rd city. When will Tesla deploy their first robotaxi?

If they take 2 yrs 11 months - that would be less than 3 years. But not by much. But, if they add the next city in 3 months - thats a big improvement.

Its not just the idea that it will take decreasing number of years between cities that is important - but also the pace of decrease.

Yes, I agree that the pace of decrease is important. But there are many factors that would affect that rate. It's really hard to even make a guess.

Waymo has NOT shown that they can scale their technology. Full stop.

Not yet. But that does not necessarily mean that they never will. Tesla has not shown that they can do L4 but that doesn't mean that they will never do L4.

 

[EVNow]

Not yet. But that does not necessarily mean that they never will. Tesla has not shown that they can do L4 but that doesn't mean that they will never do L4.

Exactly. As I keep saying, there are two dimensions to FSD. Features and geography. Waymo is ahead in features but Tesla is ahead in geography. Nobody really knows weather Tesla can get to feature parity faster or Waymo can get to geographic parity faster.

 

[diplomat33]

Exactly. As I keep saying, there are two dimensions to FSD. Features and geography. Waymo is ahead in features but Tesla is ahead in geography. Nobody really knows weather Tesla can get to feature parity faster or Waymo can get to geographic parity faster.

Agreed.

 

[daniel]

As near as I can tell from Google maps, Chandler, though an incorporated city, is really just a chunk of greater Phoenix. How useful is it really to have a (robo)taxi service that only serves Chandler (or a portion thereof)? Will Waymo serve all of San Francisco, or just a portion? For a taxi or robotaxi service to be useful it needs to be capable of serving a whole city. Not just the small part with perfect roads and uniformly good signage. Why are they opening in S.F. when they still cannot even cover all of Phoenix?

I see present robotaxi services as development platforms, not serious public services. They are a step in the right direction, but they are light-years away from competing with taxis and ride-shares, much less getting people to abandon their private cars.

Even something as simple as being able to leave your dog in the car or make multiple shopping stops without having to take all your purchases from earlier stops into each store will insure that few people will switch away from private cars.

 

[EVNow]

Even something as simple as being able to leave your dog in the car or make multiple shopping stops without having to take all your purchases from earlier stops into each store will insure that few people will switch away from private cars.

Not to speak of car seats for kids.

Why are they opening in S.F. when they still cannot even cover all of Phoenix?

Very good question. Any serious company would have tried to cover more and more areas of Pheonix with each passing quarter. Pretty terrible top management - no wonder they were all fired.

AFAIK, even in SF, they just operate in a small grid, much like Chandler.