And driving the entire LA area is just like driving in a burb of Phoenix!!
-
Want to remove ads? Register an account and login to see fewer ads, and become a Supporting Member to remove almost all ads.
Tesla, TSLA & the Investment World: the Perpetual Investors' Roundtable
- Thread starter AudubonB
- Start date
And driving the entire LA area is just like driving in a burb of Phoenix!!
The problem is that not everyone, perhaps most (no way to tell at this point), lives in the geofenced area. So someone who doesn't live there orders a robotaxi and gets refused. Not only do they never call again, but they tell their acquaintances not to waste their time. Given how long it takes to get a geofenced area certified (for lack of a better term), the first ones will only travel in a very limited area (nothing like city-wide in any reasonably large city). It will take a long time to recover from the bad publicity.
Alphacrux
Member
I have been bouncing back and forth through the 10-Q for several hours. It all makes sense pretty much; a lot of one-time items which is why Kirkhorn said it was a really complicated quarter.
I think they're effectively getting 28% gross margins and I expect it to continue for a while.
They're fine on cash as long as they deliver enough cars. (They have access to a LOT of unused credit, at least $640 million in unrestricted credit, so even if weird stuff happens like accounts payable all being due and accounts receivable all being late, they'll be fine on cash. That doesn't include things like the warehouse line.)
I expect that the more cars they deliver, the more cash they will have available for use in *building up the inventory pipeline*. That is, if they aren't delivering enough cars, we can expect another jump back to all California deliveries at the end of the quarter; if they have generated enough cash to deploy it on the pipeline, they'll keep filling the inventory in transit. Therefore I don't anticipate cash going up much -- probably just enough to make the "op cash flow - cap ex > 0" claim.
They need to deliver enough cars. So it ends up being all about production rates. Nothing new really...
FWIW I think 100,000 Model 3 delivered (while increasing inventory-in-transit) would be enough for a Q2 profit. I just don't see them pulling that off; it's higher than the likely top production rate at Fremont.
OT Cap PW/PW
If your sources are in phase, you can cut in/ cut out with no shorting.
Powerpack is not made to connect/ disconnect, it is purely grid tied.
Powerwall does islanding via the gateway module and its 200 Amp contactor.
The drop outs on switchover likely stem from a few factors:
The first is that the PW cannot control the house power supply until after the contactor has disengaged from the grid. This is a mechanical system, so there is delay there. The contactor will not start to disengage until the grid was been at the wrong frequency/ voltage for some amount of time. Put together, the house will be an non-standard voltage/frequency for some time before the PW can correct it.
The second is the speed at which the PW can go from charging to discharging. This is dependent on the design of the hardware (filtering) and software (control loop speed).
The third is stabilization time for the PW. Once the full house load is placed on the PW, it needs to match its output to the current load.
Stacking all these together and you can have a few cycles of predisconnect brownout followed by a few cycles of post disconnect brownout. If the PW was in discharge mode, the switchover should go smoother.
On top of that, if you also have solar generation, those inverters will be creating more headaches for the PW as it goes from sinking the solar capacity to sourcing the house power if the inverters cut out.
To the original question, adding large value caps to the PW high voltage rail (assuming it's topology has one) would allow supplying power while the battery side reverses direction. That would reduce one of the delay times.
If your sources are in phase, you can cut in/ cut out with no shorting.
Powerpack is not made to connect/ disconnect, it is purely grid tied.
Powerwall does islanding via the gateway module and its 200 Amp contactor.
The drop outs on switchover likely stem from a few factors:
The first is that the PW cannot control the house power supply until after the contactor has disengaged from the grid. This is a mechanical system, so there is delay there. The contactor will not start to disengage until the grid was been at the wrong frequency/ voltage for some amount of time. Put together, the house will be an non-standard voltage/frequency for some time before the PW can correct it.
The second is the speed at which the PW can go from charging to discharging. This is dependent on the design of the hardware (filtering) and software (control loop speed).
The third is stabilization time for the PW. Once the full house load is placed on the PW, it needs to match its output to the current load.
Stacking all these together and you can have a few cycles of predisconnect brownout followed by a few cycles of post disconnect brownout. If the PW was in discharge mode, the switchover should go smoother.
On top of that, if you also have solar generation, those inverters will be creating more headaches for the PW as it goes from sinking the solar capacity to sourcing the house power if the inverters cut out.
To the original question, adding large value caps to the PW high voltage rail (assuming it's topology has one) would allow supplying power while the battery side reverses direction. That would reduce one of the delay times.
Do elaborate then, as apparently this knowledge you possess deserves to be spread.
Level 5, being a drunk child in the back seat can be driven home from a school dance, requires first software and then legislative approval. It all comes down to that approval. Before than moment, no robo taxi for you.
We can distrust LIDAR because our deity tells us to. But we don't know if that's the one card the multi-billion dollar startups focused on nothing but FSD hold. Even if they do say so.
I'm not tech expert, but it tend to shink quickly. My phone has a pretty awesome camera and I can remember bulky digital cams that took a 3.5" diskette. Funded startups have no incentive to show their hand early, look at Rivian.
It would be interesting to know how much time of their overall driving at Waymo is spent autonomously.
On Saturday I was behind one of them in Palo Alto, it went to make a left turn, then stopped, halfway into the junction when the lights changed. Other cars had to drive around it. Then when the lights changed it made the left quickly and held up the opposing cars. I could clearly see the driver's hands on the wheel as it went. So was that a disengage? or was it simply gathering data? or was someone using the car to run an errand?
On Saturday I was behind one of them in Palo Alto, it went to make a left turn, then stopped, halfway into the junction when the lights changed. Other cars had to drive around it. Then when the lights changed it made the left quickly and held up the opposing cars. I could clearly see the driver's hands on the wheel as it went. So was that a disengage? or was it simply gathering data? or was someone using the car to run an errand?
EVNow
Well-Known Member
So, they don't have to report if the car sits at a junction and refuses to take a left turn because of the oncoming traffic ?
- Disengagement rates are a nonsense metric. If a human driver voluntarily takes over because the car is approaching a scenario it is not programmed for, this does not have to count as a disengagement. The car can also be programmed to take the long easy route, and can still be way too hesitant with its actions. Tesla is not trying to optimise for a meaningless KPI, it is trying to put in place a platform to identify, prioritise and quickly fix generalised scenarios and edge cases one by one. Even in Phoenix alone, Waymo will hit a brick wall when they fix all the edge cases in their 10 million miles of driving (if this is even possible without orders of magnitude more data), but they will have no idea on the millions of edge cases which happen infrequently and did not occur in their 10 million miles.
- Waymo's map, model, simulate and hand code every turning and junction strategy is unscalable. Solving Phoenix (also where there are no hills or rain) has only minor transferable capability to new regions. They are spending $1bn+ per year on solving Phoenix alone.
- All Waymo cars are constantly monitored from HQ sometimes with override from HQ. This again is not scalable and not a real solution.
Last edited:
If they will leave the car sitting in the middle of the road then yes. But if they want to proceed driving with it then they need to disengage and driver makes the turn.
I am not sure if they are able to change the destination so the car will go another route if there is a lot of traffic in an unprotected left turn.
RobStark
Well-Known Member
Silevo tech was about 2 years ahead of the current tech when the merger happened.
Tesla delayed delayed manufacturing panels.
At this point it has been superseded.
EVNow
Well-Known Member
But would they then report it ? Because one could argue the FSD didn't fail, but was just taking too long, and there was no danger.
Waymo was at 0.64 disengagements per 1000 miles in 2015. Tesla is at best 10 disengagements per 1000 miles according to the investor day. And Waymo is still not done with the fully self driving product even in Phoenix. How could Tesla go from 10 disengagements per thousand miles to 0 in one year or even 2-4 years when Waymo has taken 4 years to go from 0.64 disengagements to 0.09 disengagements.
PS. I know that Waymo is geofenced. For customer sitting in a Robotaxi it is not important if it is geofenced or not.
You're answering your own question!
Look at the 15/16 period: down to 30% as many disengagements with only 50% more miles. There's clearly more to it than just collecting miles...
Look at the 17/18 period: 50% improvement with 3.5 times as many miles. Maybe it scales with collected miles as well?
Tesla certainly has the miles.
Why do you deny them the possibility to also have the brains to do non linear improvements?
Last edited:
It failed because it got stuck. If it is immobile in the middle of the road then it is a safety hazard to all other cars sharing the road.
I am not sure why we need to try to *sugar* on other competitors who have the same aims. For me as an investor:
1) If Waymo and GM Cruise are approved for Robotaxis this year then it will automatically probably increase price of TSLA shares 30-40% because people see it is actually a viable technology and all the frontrunners will benefit.
2) If Audi would make this year 50% of the sales from their EV-s then the price of TSLA would increase again massively because they are the frontrunner for EVs.
I hope that both GM Cruise and Waymo succeed as soon as possible to increase the value of TSLA. And I hope that Etron and iPace will sell like hot cakes because it means that EVs are finally taking off like they should.
I hope they all succeed too. Just pointing out that their current strategy is stupid.
EVNow
Well-Known Member
I'm fairly sure in both the cases TSLA SP won't go up. Enough FUD will be created to say Tesla is losing against competitors.
I think it is realistic that the Robotaxis will start on Airport - City centre routes or something like that. It will get people used to the concept and help to make the regulations. This would be beneficial for TSLA. Don't have to be the first to the market but the best.
Similar threads
- Locked
- Locked
- Locked
