Wild idea: Tesla gives up on autopilot for Model S, offers Model X with autopilot to anybody with Model S that has autopilot hardware and appropriate package that includes it, and takes back those Model S's with a superb trade in value to be resold CPO with advertising to used buyers that they will never have autopilot, and the marketplace gets a bunch of depreciated used Tesla cars on the road in the hands of happy poorer buyers that will take up some of the Model 3 demand, smoothing out the overall marketplace of Tesla model availability.
-
Want to remove ads? Register an account and login to see fewer ads, and become a Supporting Member to remove almost all ads.
Wild idea: Tesla gives up on autopilot for Model S, offers Model X with autopilot to anybody with Model S that has autopilot hardware and appropriate package that includes it, and takes back those Model S's with a superb trade in value to be resold CPO with advertising to used buyers that they will never have autopilot, and the marketplace gets a bunch of depreciated used Tesla cars on the road in the hands of happy poorer buyers that will take up some of the Model 3 demand, smoothing out the overall marketplace of Tesla model availability.
I feel like a broken record, but people still keep doubting Tesla can use the current sensor suite to deliver [most of] what they promised?
Tesla is doing AP beta testing now. Which implies that AP is coming out. It will not work in 100% of situations. Just like TACC doesn't work in 100% of situations.
This is not autonomous driving (Level 3). Considering that the driver needs to be fully aware of the surroundings, even if AP works in 99.9999% situations, they can call it a success.
Self parking (parallel) is easy given the sensor suite. You pull up to a parking space, hit the park button, the car tells you to line up your car with the empty box on the screen, you click go. Once you click go, it's simple geometry. Turn steering wheel one way to a certain angle, pull back a certain distance, turn the other way at a certain angle, keep pulling back a certain distance. Tada you're done.
Stall parking is going to be harder without 360 degee camera or side cameras. But given the AP has a front and rear camera, you can see the stall, click park, and have the car back into it. Also, stall parking may be accomplished by you (the driver) making the first turn to line up the rear camera at least somewhat with the parking spot (line up the box in the rear camera), and have the car finish off the parking job.
I still think most of it is achievable. It might not be as autonmous as the forum members hope, but I don't see any real limitations in the current hardware. People keep thinking that the self parking is going to be some magic button you hit from 17 miles away going sideways, and the car will magically flip you around, steal the parking space from the other person trying to get in it, and park you perfectly 100% of the time. No, that wont happen. It'll assist you, just like AP will assist you in driving. None of it is autonomous.
/soap box
Tesla is doing AP beta testing now. Which implies that AP is coming out. It will not work in 100% of situations. Just like TACC doesn't work in 100% of situations.
This is not autonomous driving (Level 3). Considering that the driver needs to be fully aware of the surroundings, even if AP works in 99.9999% situations, they can call it a success.
Self parking (parallel) is easy given the sensor suite. You pull up to a parking space, hit the park button, the car tells you to line up your car with the empty box on the screen, you click go. Once you click go, it's simple geometry. Turn steering wheel one way to a certain angle, pull back a certain distance, turn the other way at a certain angle, keep pulling back a certain distance. Tada you're done.
Stall parking is going to be harder without 360 degee camera or side cameras. But given the AP has a front and rear camera, you can see the stall, click park, and have the car back into it. Also, stall parking may be accomplished by you (the driver) making the first turn to line up the rear camera at least somewhat with the parking spot (line up the box in the rear camera), and have the car finish off the parking job.
I still think most of it is achievable. It might not be as autonmous as the forum members hope, but I don't see any real limitations in the current hardware. People keep thinking that the self parking is going to be some magic button you hit from 17 miles away going sideways, and the car will magically flip you around, steal the parking space from the other person trying to get in it, and park you perfectly 100% of the time. No, that wont happen. It'll assist you, just like AP will assist you in driving. None of it is autonomous.
/soap box
gizmoboy
Member
#4 falls into the workaround category; dumping itunes library onto USB stick. I think I can dump my audible books as well (haven't explored that as thoroughly as yet, but IIRC, I can also download them into iTunes and then dump them out)
I can speculate that Model S will get following updates
- New Autopilot hardware (2.0) with some features available immediately and some with software updates
- 90KW will be standard (no more 85KW)
- New processor
- 20" wheel option
- Wiring for future retrofit of side mirrors to cameras
- Updated interior:
- Ventilated seats
- Adjustable seatbelt height
- USB port for back seats
- Center console will be standard
- New options of textiles and trims
- Updated exterior:
- New nose cone design
- LED headlamps
- New exterior colors
MarkS22
Member
I've said it before, but I don't see any significant Autopilot upgrades for several years. Incremental Autopilot 1.1, sure. (Things like 360 camera views, slight improvements like more places to self-park.) But the next leap happens when MobileEye EyeQ4 comes out. The EyeQ3 just came out late last year when Autopilot was announced. The engineering samples for EyeQ4 aren't even out yet, with production expected in about two years. This isn't guessing, but actual road maps from publicly traded companies.
EDIT: The EyeQ4 will allow for 8 camera views, including a cluster of 3 forward (including a wide angle and a detailed narrow FOV for lane markings and small road debris), two cameras on each side, and one rear along with ultrasonic and radar to have a complete, detailed view of the entire car's surroundings. That's what we need for the next level of Autopilot, which I'd call "2.0" and it'll be here eventually, but we're looking at 2017-2018.
Last edited:
Urgo
Member
Ahh.. yes.. was hopeful for a moment. Thought you had some aftermarket solution that put in an aux in port or something and got excited haha.
I think you will find yourself to be so wrong about this prediction before long.
Even a high end Ford these days has a better sensor array than a current Model S. The Ford for example has sideways radar that can detect a car 15 yards away as you back out of a driveway or parking spot. To put how limited the Tesla sensors are in perspective, a Ford Taurus (a basic economy car) will know of a car approaching it from the side that is 45 feet away and a Tesla would only know of that car when it is 16 feet away with the current sensor array.
Why are you so fixated on MobileEye and their schedule? Tesla might go for a hybrid sensor array from a variety of different sources as they are a tier-1 integrator of this technology and they do their own software. Tesla might end up going with Drive PX or they might go with a Drive PX, MobileEye, hybrid setup, or they might decide to build their own system. Look at the sensor array of a current S Class. It has stereo vision and multiple redundant levels of long, medium, and short range radar and they are not limited by some arbitrary MobileEye schedule. Why should Tesla be?
I expect Autopilot Ver. 2.0 sensors to be released with the Model X and it will be available in the Model S before long. You can call it ver. 1.1 if it makes you feel better but expect to see a much more robust and capable sensor array in the Model X, and the Model S. Arguably the current sensor array of the Model S is the most limited sensor array in the industry in terms of range and coverage while it promises just about more than anyone else in terms of functionality.
They will delivery basic Autopilot functionality with the existing Ver. 1.0 sensor array but I expect a much more robust sensor array to be released soon.
Gizmotoy
Active Member
At some point it's an argument over version numbers, but I agree it doesn't make a ton of sense to say that they're definitively tied to the MobileEye schedule. Even today, as you note, the sensors are limited compared to basic systems on other vehicles. I mentioned it in another thread, but the Model S has rear ultrasonics instead of the millimeter wave radar for blind spot detection that everyone else is using. They're slower and have substantially reduced range. I wouldn't be surprised if the Model X launch brought a sensor revision with what they've learned so far. If not to add new features (the low hanging fruit here are improved blind spot detection and indication and the mentioned cross-traffic detection), than to improve performance and reliability.I think you will find yourself to be so wrong about this prediction before long.
Even a high end Ford these days has a better sensor array than a current Model S. The Ford for example has sideways radar that can detect a car 15 yards away as you back out of a driveway or parking spot. To put how limited the Tesla sensors are in perspective, a Ford Taurus (a basic economy car) will know of a car approaching it from the side that is 45 feet away and a Tesla would only know of that car when it is 16 feet away with the current sensor array.
Why are you so fixated on MobileEye and their schedule? Tesla might go for a hybrid sensor array from a variety of different sources as they are a tier-1 integrator of this technology and they do their own software. Tesla might end up going with Drive PX or they might go with a Drive PX, MobileEye, hybrid setup, or they might decide to build their own system. Look at the sensor array of a current S Class. It has stereo vision and multiple redundant levels of long, medium, and short range radar and they are not limited by some arbitrary MobileEye schedule. Why should Tesla be?
I expect Autopilot Ver. 2.0 sensors to be released with the Model X and it will be available in the Model S before long. You can call it ver. 1.1 if it makes you feel better but expect to see a much more robust and capable sensor array in the Model X, and the Model S. Arguably the current sensor array of the Model S is the most limited sensor array in the industry in terms of range and coverage while it promises just about more than anyone else in terms of functionality.
They will delivery basic Autopilot functionality with the existing Ver. 1.0 sensor array but I expect a much more robust sensor array to be released soon.
“The X will have a step beyond what we’re currently shipping on the autonomous front,” Musk said. “There are a whole bunch of things about the X that have not been revealed.”
That's about all they mention about it in that article...but it does hint to something more than what the S currently has enabled.
That's about all they mention about it in that article...but it does hint to something more than what the S currently has enabled.
gizmoboy
Member
Soolim
Member
MX is overkill for my needs. No deal. I speculate it will cost no more than $1,000 extra (Tesla internal cost not consumer price) to get the proper sensors to make the MS version 1 AP work.
- - - Updated - - -
Firstly, we don't know to what extend is the AP beta. Is it just the lane keeping that is being tested? Beta test by EAP is unlikely to include self parking, because the test environment can easily be created right at Tesla parking lot. Just get a few staff to try parking the car is easy enough.
You said stall parking is harder without 360 degree camera, and that is my concern and why I keep saying that present sensors are deficient. Lining up the rear camera with the parking spot correctly is 75% of the job. That should be the job of the AP, not the driver, otherwise why bother?
MarkS22
Member
I think we're just defining the word "significant" different. I owned a 2013 Ford Fusion with BLIS. I'm very familiar with the technology. I've been following it for quite some time. My 2002 Infiniti QX4--13 years ago--had radar-based Intelligent Cruise Control.
I've never doubted incremental upgrades. Tesla is known for constantly refining technologies as they go. I'm specifically doubting a major next level of autonomous driving. I think the next things will see will be enhanced ultrasonic sensors, surround view, maybe even a second camera (or a better camera) up front to help with their problems on lane keeping (with white roads and white lines). I just don't see a brand new brain controlling it all.
I don't think the MobileEye schedule is arbitrary. The EyeQ3 can already support new features and some sensor upgrades. I don't think Tesla has the same level of resources to simultaneously work on getting the EyeQ3 working for Autopilot 1.0 while supporting a completely new sensor array and processor. Whether it happens on the X reveal or not, I suspect the next round of Autopilot hardware will be ways to make the EyeQ3 work better, after having tested it for the past year. They'll address all the quirks and "corner cases." EyeQ3 has support for "Surround View," so I see that happening. I also see the addition of pedestrian detection and LED headlamp control (i.e. following the curve of the road). Now, I consider those "1.1" features... not "2.0" but it's semantics. Maybe we're already thinking the same thing.
Simply, they have some way to go before even maxing out the current system already being installed, so I think we'll see that first. If they add a lot more (or new processors)... awesome. I will be very pleasantly surprised.
I gave a sample use case, I could be 100% wrong and Tesla created a more sophisticated approach by... whatever. I could also be 100% wrong, and they could just not give stall parking due to lack of hardware. Though I really do believe it's doable, and that it wont be as pretty as everyone on the forums hopes.
Question: What's the big deal about AP and stall parking? I understand it was promised, that taken aside. How hard is it to stall park? I feel like it would be faster for me to to park the car in a stall myself, then line up the car, hit a button and wait for it to slowly roll back in.
To be fair, same for parallel parking FOR ME. A lot of people don't know how to parallel park, so I can see where they want the car to do it for them.
I'd say most Model X features will make it to the Model S within 6 months, as the Model S is their bread and butter and it would be horrible if potential Model S buyers sit on the sidelines before the Model X goodies make it to the Model S.
I bet already a bunch of potential Model S customers are waiting on the sidelines for the Model X launch and the subsequent enhancements to the Model S.
While they ramp up production of the Model X (which is likely to take months), they can't afford to have a slowdown of Model S sales so once they get the MX out of the door I'm sure they will prioritize transferring the new technology to the MS. As an added benefit, it simplifies their supply chain and production the more technology and parts the two vehicles share.
Urgo
Member
Sort of.. I was waiting about a month until the design studio went live to place the order. My order confirms today and should be delivered around the time/shortly after the X's start coming out so was hoping anything that's coming soon they might just add to my S90D as well. If it's going to be six months oh well will get those features in my next car.
Soolim
Member
Since it is promised, then it should be delivered. I can see the challenges in lane keeping as it deals with road pavement marking, weather condition, and the danger to life and safety, so I am willing to give them some slack. I keep asking myself, if the sensors are sufficient, and software update by OTA is the practice, why can't Tesla not incrementally release the AP with the self parking first? Would you not agree that self parking should be a piece of cake for Tesla (since it is older technology), and after 10 months, unless they encountered some unforeseen H/W difficulty?
Similar threads
