MarkS22
Member
Yeah, I know you're on the sidelines waiting. I do think your wait is under a year. You're a lot closer than when (if I remember correctly) we discussed this before the X launched.
-
Want to remove ads? Register an account and login to see fewer ads, and become a Supporting Member to remove almost all ads.
Why AP 2.0 Won't Be Here Soon, and It Won't Be What You Think It Is
- Thread starter Todd Burch
- Start date
-
- Tags
- Model S
Yeah, I know you're on the sidelines waiting. I do think your wait is under a year. You're a lot closer than when (if I remember correctly) we discussed this before the X launched.
You must not have looked up solar city burn rate and that added to teslas... Anyone would be desperate to post good numbers to be able to raise decent capital... You know to stay in business and not go bankrupt.
Big picture.
technomation
Member
AP is not just a novelty feature any more; it's become a huge differentiator on par with EV battery range. Granted, Tesla was clever with their recent OS 8.0 software update, but by Elon's own admission, what they did was facilitated mainly by new radar-interpretation algorithms given to them courtesy of Bosch engineers who have far more domain knowledge in radar. Meanwhile the German automakers--who rule this planet's luxury auto market--have gotten over their earlier hubris and are now acknowledging Tesla's wild success with EVs and with autonomous driver assistance.
If Tesla is asleep at the wheel on what Mercedes and Audi have recently accomplished with autonomous capability, then it will quickly fall behind and lose the autonomous game within the next 12-18 months. Clearly Tesla has to move forward decisively with a new AP 2.0 hardware platform, so that they can continue developing higher-level autonomous capabilities with software not limited by the current obsolescing AP 1.0 platform.
Last edited:
I certainly agree that MB lacks the fleet learning infrastructure to really do a whole lot with those sensors.
The problem is you need BOTH sensor fusion, and the fleet learning combined.
Truck lust wouldn't occur on an MB because the car would be able to better detect the semi truck/trailer, and wouldn't hunt into the truck while it was searching for the line. Sure Tesla can probably tweak it a bit, but it's still not the same.
With the current Autopilot package the Tesla can't really do collision avoidance that effectively since ultrasonic sensors can't really be used for side monitoring. We all know the side monitoring of the Tesla is horrible.
Tesla needs more Sensors. They're setting themselves up for false positives if a single sensor can activate AEB (all the way to a stop). Imagine if there is something wrong with the sensor or it was a false detection.
MB needs to move to fleeting learning, and connectivity. But, more importantly they need a bit more bravado. If you look at the MB they have excellent crash prevention, and lots of videos demonstrating it. But, their AP system is woefully limited. What good is a super safe car if you don't want to drive it?
Personally I like Volvo's sensor suite (which includes Lidar), but they just don't have the software. They also don't score that well in any of the tests that I've seen.
Last edited:
Canuck
Well-Known Member
I predicted it would come out before the end of 2015... ha! That sure turned out to be wishful thinking since I was really wanting to upgrade before the fatality and before I had an AP loaner for a weekend.
Now I've decided to keep my large frunk, 80 amp, "classic" that I still enjoy driving every day and adding a loaded Model 3 to zip around in and replace our Leaf. So Tesla can take their time as long as the AP2.0 hardware is on the Model 3, but that means they need to get it on the S/X first, so hence my bias still keeps pushing for it soon.
Welcome to the club!
Tesla's development team likely already knows what sensors they want to have in AP 2.0 - and have been testing the hardware.
The AP 1.0 improvements in 8.0 were unexpected, because they are able to do more than they originally anticipated with the current hardware - and hopefully that will be enough to provide assurances to the AP 1.0 owners, prospective owners, governments, and shareholders. But there's a limit to the improvements they can make with the current sensors and the very limited processing power.
Tesla needs to get the AP 2.0 hardware out - and it's unlikely they would build Model 3 with the "old" AP hardware and then quickly plan to replace that with the AP 2.0 hardware.
While it may take a year or longer for Tesla to develop the software needed to fully utilize the AP 2.0 hardware, it still seems likely they'll ship AP 2.0 on Model S/X at least 3 to 6 months before they begin Model 3 production.
And, because Tesla doesn't want to kill sales, it's also very likely we won't find out about AP 2.0, until they've already started put the new hardware in production, so there'll be a small group of new owners who will be pleasantly surprised that their cars get AP 2.0 when they expected 1.0.
Still planning to stick to our strategy of waiting until 100D and AP 2.0 is announced before we order our next Model S.
The AP 1.0 improvements in 8.0 were unexpected, because they are able to do more than they originally anticipated with the current hardware - and hopefully that will be enough to provide assurances to the AP 1.0 owners, prospective owners, governments, and shareholders. But there's a limit to the improvements they can make with the current sensors and the very limited processing power.
Tesla needs to get the AP 2.0 hardware out - and it's unlikely they would build Model 3 with the "old" AP hardware and then quickly plan to replace that with the AP 2.0 hardware.
While it may take a year or longer for Tesla to develop the software needed to fully utilize the AP 2.0 hardware, it still seems likely they'll ship AP 2.0 on Model S/X at least 3 to 6 months before they begin Model 3 production.
And, because Tesla doesn't want to kill sales, it's also very likely we won't find out about AP 2.0, until they've already started put the new hardware in production, so there'll be a small group of new owners who will be pleasantly surprised that their cars get AP 2.0 when they expected 1.0.
Still planning to stick to our strategy of waiting until 100D and AP 2.0 is announced before we order our next Model S.
I agree. Elon has stated that the S/X will always get the cool new features first before any lower priced Tesla models. In my opinion, AP 2.0 obviously has to come out on the S/X many months before the 3 is available. That will spur S/X demand at a time when many people will be pondering whether to place a 3 order and wait a long time or pay more and get a Tesla within weeks, not years. For those who simply cannot afford the higher price of the S/X, the release of AP 2.0 will heighten their excitement over their Model 3 reservation and decrease the chance of them cancelling and buying a competitor's car.
I think I agree with your reasoning in principle. I'm not sure I would have said many, though.
I'm thinking this logic should point to AP2 on the S/X 3-4 months before, right when the orders might slump because folks are willing to wait for the 3.
( This might be what you meant by many,I suppose, but I was reading it as an argument for release this year when you said many months.)
I agree, IMO, the current AP 1.0 is a liability and a architecture dead-end. The improvements in 8.0 were necessary to improve the safety due to recent accidents that have occurred.
Tesla probably already knows what sensors and cameras they need, but with the Mobileye divorce, they need to find new silicon to run and process all the data. Unfortunately that takes time, especially if they are going to roll their own design vs using something more general purpose computing related like an nVidia solution.
technomation
Member
Tying the introduction of AP 2.0 to that of the M3 is still highly uncertain. Green Car Reports just published this poll of "Elon Time" today:
When do you think the first Tesla Model 3 electric cars will be delivered?
34% Jul-Dec 2017 (per Tesla)
38% Jan-Jun 2018
20% Jul-Dec 2018
8% 2019 or later, or never
http://www.greencarreports.com/news/1106250_when-do-you-think-the-first-tesla-model-3-electric-cars-will-be-delivered-poll-results
So based on a random poll of folks visiting an EV enthusiast site, most (all?) of whom presumably have no knowledge of the actual schedule, critical path, or program status, we're supposed to conclude what, exactly?
RogerHScott
Active Member
The length of the emperor's nose, to four significant figures, of course!
technomation
Member
You might conclude that we on this enthusiast site, who have no more insight than anyone else, are nonetheless aware of the Tesla foible of "Elon Time" (e.g., after making a down payment for Model X, I waited over two years, but had to give it up when there was still no delivery in sight). This poll simply corroborates what everyone knows--that Tesla usually misses its projections, so any attempt at predicting when AP 2.0 will debut, dependent on the Model 3 schedule, is ABSOLUTELY FUTILE
3Victoria
Active Member
1. Bosch supplied only a new driver that let Tesla control and query the radar unit more fully. The algorithms are all Tesla's.
2. AP2 hardware is coming, perhaps sooner than we think.
3. I guess Model 3 reveal deux will be Mar 31 2017 ... one year anniversary from first.
Sweet dreams one and all!
2. AP2 hardware is coming, perhaps sooner than we think.
3. I guess Model 3 reveal deux will be Mar 31 2017 ... one year anniversary from first.
Sweet dreams one and all!
Canuck
Well-Known Member
Does anyone know what is happening to the Mobileye EyeQ3 processor being used by Tesla for AP now that it has broke ties with Tesla? Are different chips going in the cars now? If not, did they break ties but still use the Mobileye's hardware? Given how Mobileye is bashing Tesla, that seems odd, if true.
Sure. Disassemble the mirror housing on an S or dig behind the glovebox on an X, then read the labelling on the AP module. My bet is that all cars to date continue to have an EyeQ3 and they still have a contract until some future date (end of the year?), but I could be wrong.
Canuck
Well-Known Member
That's my guess too but for how much longer is the question? Those who are predicting no changes for some time perhaps didn't consider this? We know a faster processor is needed, so perhaps Tesla will start there? The more I think about this, the more I think we will see new hardware soon -- not that I know anything -- just wondering out loud here.
Similar threads
