I like your light heartedness. But if you spend $15k for a puppy it better win every race.
-
Want to remove ads? Register an account and login to see fewer ads, and become a Supporting Member to remove almost all ads.
Waymo vs FSD 11.3.6
- Thread starter Porkalletta
- Start date
How does Waymo respond and work with new construction sites? That's were I have to biggest problem with my Tesla. Construction is non-stop and in different locations here in SoCal all the time.
Some people are still drinking the Kool Aid. Not sure of their motives... maybe they just don't know any better, maybe they emotionally attached, maybe they're financially attached (TSLA).
For me, the "comparison" was over at the very beginning when the Waymo car pulled up without a driver inside.
My opinion is that Tesla's sensors aren't capable of delivering level 5 autonomous driving, only level 2 forever. Cameras are not adequate... LiDAR is required. Also, Waymo engineers said long ago that Tesla has hit a "glass ceiling" and I believe it.
For me, the "comparison" was over at the very beginning when the Waymo car pulled up without a driver inside.
My opinion is that Tesla's sensors aren't capable of delivering level 5 autonomous driving, only level 2 forever. Cameras are not adequate... LiDAR is required. Also, Waymo engineers said long ago that Tesla has hit a "glass ceiling" and I believe it.
I guess we should view it as a good driver assist program and that's what we are really paying for. Not future promises.
Tesla is way mo ahead. Remember Musk’s LA to NYC Tesla FSD trip in December 2017? As you should recall, the Tesla was fully autonomous and charges at Superchargers were performed without leaving the car to initiate.
Autonomy was solved by Tesla in 2017 just as Elon proclaimed. Why these “status report” threads keep popping up six years later is a mystery.
Autonomy was solved by Tesla in 2017 just as Elon proclaimed. Why these “status report” threads keep popping up six years later is a mystery.
It was Mobile Eye and they said the camera based system will likely "get into a glass ceiling", which hasn't happened yet, but with the current camera placement, I tend to agree.
Driverless helps a bit but the big cost is Waymo's sophisticated low production system. Performance, redundancy, and safety today. Optimization and profits later.
I guess we should view it as a good driver assist program and that's what we are really paying for. Not future promises.
I think the promises were vague, even made before the "official" autonomous driving levels were in place. I wasn't trying to throw fuel onto a fire either... sorry if it seemed that way.
I suspect that Tesla will stay the course indefinitely, because they took a lot of money from people based on future performance... but without a deadline.
Tesla doesn't even do a great job with *regular* predictable software where the user just hits buttons and it has to do something simple (like turn on the heated seats).
FSD isn't going to get vastly better than it is today unless something significant changes. I'm not saying that's impossible... generative AI is mind blowing and relatively in it's infancy. So perhaps Tesla will stumble on a much better path forward.
EVNow
Well-Known Member
I like your light heartedness. But if you spend $15k for a puppy it better win every race.
So you are saying as long as FSD "wins" some "race" you are happy having spent $15k (apparently without doing any homework) ?
So you are saying as long as FSD "wins" some "race" you are happy having spent $15k (apparently without doing any homework) ?
That's how it works for people at horse races or roulette tables, right?
Not saying FSDB is the winner or anything, but this drive demonstrates that Waymo is the real loser. I can't imagine in any scenario in which a waymo at it's current form is better than a Uber. It takes roughly 2x the amount of time to arrive at its destinations due to the conservative routes it picks. This not only increase the operating cost of the Waymo beyond the cost it is suppose to offset from a real human driver, but also no real person besides using it as a novelty would want to spend similar vs the uber to sit in a metal box 2 times longer.
So until Waymo start not only work on scale but also its ability to take normal routes that's faster than 45mph, this product is DOA. So far I am still in the camp that it's either Tesla's FSD or bust as a profitable robotaxi business.
What would prevent Waymo from scaling up the number of vehicles?
In one corner, we have a company who already operates actual functional robotaxis, albeit they don't yet drive on the thruway.
In the other corner, we have a level 2 ADAS that puts all liability on the *HUMAN DRIVER*, uses regular cameras which are frequently blinded by all different types of external factors, has many known limitations, frequent disengagements which require the driver to take over in a moments notice, massive amounts of videos where the ADAS caused or almost caused an accident, and several fatalities... and you think *THAT* technology is in the lead because you paid a bunch of money for it?
They are two completely different approaches and end goals. Waymo doesn't ever intend on achieving Level 5. Their CEO said it's not possible. Tesla believes they can, but regardless of the probability of Elon's beliefs, Tesla wants it to work everywhere instead of specific areas/conditions.
EVNow
Well-Known Member
But this is neither... It's like buying an XBox and feeling good or bad based on some benchmarks with playstation a website publishes.
Well Waymo is not a car manufacture for one. Tesla will probably slap 8 cameras on the cheapest giga printed platform they can develop and just mint them in the millions per year. Waymo has to partner up, which means a lot of bureaucracy, installing 6 figure equipements on the cars and testing them first before allowing them to become robotaxies. Not to mention legacy auto would want to profit share as they are sharing liability.
This is just one part of it. The other part is their geofence nature which has high operating cost. If 2 cities and 1200 waymo cars lose them 2B a year, how many Billions will they lose a year if this was 200 cities and 12 million waymo cars? There's not enough money in the world that can support that type of scale. This is why I would rather see them demonstrate a positive operating margin first before expanding so then expanding doesn't = they will just incur more and more losses.
Last edited:
diplomat33
Average guy who loves autonomous vehicles
In its current form, you would be correct. But Waymo will not stay in its current form. Waymo can do normal routes now and do speeds up to 65 mph. And Waymo can do highways now. Waymo is letting employees do highway routes. They just have not enabled the faster routes for the general public yet but that will happen very soon. So Waymo will do shorter routes soon. What we are seeing now is NOT the final state of Waymo's ride-hailing.
PS: Tesla FSD Beta only "wins" this comparison because Waymo has not enabled highways yet for the general public. FSD Beta would not "win" when Waymo does allow highway routes for the general public.
Last edited:
diplomat33
Average guy who loves autonomous vehicles
They will not continue to operate at a loss. Waymo is working to bring down operating costs so that they can achieve profit per ride. Then, they can scale to many cities and more cars and increase profits.
I honestly don't see how robotaxies will ever be profitable the way Waymo/Cruise are operating. The problem I see is that waymo/cruise are overly complicating a relatively simple human task. If the task was hard, then the human would be paid a lot more. However it's easy so the human cost Waymo is trying to remove is actually not all that great. So unless you simplify the system by getting rid of all the geo fencing engineers, the human testing engineers, and all the customer service personals, then how will this service ever offset the cost of a human driver?
diplomat33
Average guy who loves autonomous vehicles
Not sure what you mean by "simple human task". Are you talking about driving? Driving might seem easy for humans but it is super hard for computers. Autonomous driving is one of the most difficult problems that humans have ever attempted to solve. The fact that we even have safe and reliable driverless cars in a geofence is an engineering miracle.
How to make money as a robotaxi network.
1. Build an app
2. Gets a cut from client's profit
3. Have clients clean/fix/operate the car
4. Have clients do the customer service
5. Have clients be the car tester(fsdb)
6. Have the clients pay for FSD while having FSD equipments be very cheap to manufacture
7. Have zero geo fencing engineers
8. Scale to any city anywhere all at once
9. Have the clients pay for 100% of the equipment while you make a 25% margin
Tesla's operating cost is barely anything...whatever it cost to service the app. The rest are being pushed onto the client while they set up their robotaxi business.
1. Build an app
2. Gets a cut from client's profit
3. Have clients clean/fix/operate the car
4. Have clients do the customer service
5. Have clients be the car tester(fsdb)
6. Have the clients pay for FSD while having FSD equipments be very cheap to manufacture
7. Have zero geo fencing engineers
8. Scale to any city anywhere all at once
9. Have the clients pay for 100% of the equipment while you make a 25% margin
Tesla's operating cost is barely anything...whatever it cost to service the app. The rest are being pushed onto the client while they set up their robotaxi business.
Similar threads
