Separate names with a comma.
Discussion in 'Model S' started by Rafik, Oct 10, 2015.
At least for now
The software is the same too. Non-existent.
Why does the Model X photos on the Tesla Motors Web Site then show a stereo front camera system?
Show me one ounce of proof other than speculative extrapolation that that is in fact what they are. Even so, I did remark that these are the "current state of affairs." No one knows if and when things will change. Considering Elon has stated that further autopilot functionality ala autonomous driving is roughly 2 years away, it's difficult to imagine why they would put new hardware in this early. In the case of the autopilot V1, no one really knows whether Tesla intentionally put the hardware in early to have a larger fleet of AP capable cars at the time of roll out or if they just miscalculated needing an extra year to actually roll out the functionality. Even then, they were still able to make some functional use of the hardware before they reached that point. Other than full fledged autonomous driving, I'm not sure what more V2 hardware would bring to the table that the V1 hardware isn't already capable of. Yes, there is an argument for more sensors to enhance safety features, such as side collision avoidance, but who knows whether Tesla really thinks it's necessary at this point.
V2 hardware would presumably have more radars and allow the car to change lanes on its own, without requiring the driver to use the turn signal to initiate a lane change.
Elon said in February that the X would have autopilot that is a step beyond what the S could offer. We've seen X mules with additional AP hardware so I'm sure they will have additional features eventually.
If you wanted to nitpick, you could say that two of the same camera rather than one isn't different hardware.
I guess if you extrapolate that, as long as the Model X's sensors are limited to radar, sonar, and optical I guess you could technically say the hardware is the same
It seems the X has greater coverage in range and redundancy for the sensors.
It's also possible that the slightly additional height of the Model X makes the sensors more effective.
Re-read the question posed to Elon - it was regarding capabilities, not hardware. The Founders cars that were delivered had only one camera, but the assembly is large and looks like it was designed to accommodate two cameras. The single camera sits off to the left with a big, blank space to the right where the 2nd camera is "missing" compared with the photo on Tesla's web site. I'm pretty sure a 2nd camera is coming to the production vehicles.
Another "they are holding stuff back" post.... I don't think so. The car is awesome with current generation tech. Autopilot is just getting turned on this Thursday, and only partially. Tesla needs to focus on the gigafactory, model 3 to make it as a company.
I got to explore a founders X car -- it's just like an S, albeit with amazing doors. The S is already amazing, they don't need more right now. Why? They have Zero competition.... ZERO... For the next two years at least.
I expect wider changes in 2017 ahead of the competitive releases.
You guys are reaching. "Is there any difference between the hardware and capabilities?" Musk: "Same." And you're still arguing that a vague few pixels on the site means new hardware? The official press release for the X also specifically says "a forward facing camera" (singular). This is a company that announced Autopilot 1.0 a year ago and is just turning it on Thursday.
Of course there will eventually be upgrades. But they didn't launch with the X. Multiple sources confirmed the S and X will be in "lockstep" for Autopilot hardware and feature. It makes absolutely no sense to splinter Autopilot development. Nor does it make sense to upgrade hardware before the original is maxed out. The X was released, press releases went out, and now Musk confirms they're the "same" Autopilot hardware and capabilities... and we're still debating secret features?
Musk is the CEO of a public company. If he says "same" hardware and capabilities, it has to be. There are stock trading implications if he made a false comment. Furthermore, upgraded sensors are much less important to the target demographic of the X than the S. Case in point, browse the forums and look for complaints about X Autopilot features verses, say, folding seats. Saying "surprise! We added a second camera to see highway lines better and a rear radar to see cars approaching better!" as their ace in the hole makes no marketing sense.
While I agree with your view that its going to be the same, I did find this quite funny given his history. "Same" can be interpreted in so many ways..!
I must be missing something, because the questions clearly states "hardware and capabilities."
No one said that the X isn't going to get newer hardware at some point, and perhaps that's where it will sit. Or perhaps that space houses additional hardware and components to make the overreaching windshield possible...who knows.
Once again - "hardware and capabilities" - not quite sure why you're desperate to make something out of nothing.
Elon did say that in February, but that also never surfaced when the X launched. Tesla even went out of their way to avoid the conversation altogether. That, along with Elon's most recent comment, leads me to believe that they have sidestepped that original plan. We have seen both X mules and S mules with additional AP hardware, iirc.
The X and S AP hardware are identical for now - how soon that changes is unknown, but everything Tesla has said points to the conclusion that when the X changes, the S will change as well. It's very possible that the new hardware will be in cars by the time X sig cars come off the line...or maybe production cars. Either way, I just feel it's too premature to speculate about it. V1 is just now coming to fruition and I wouldn't be surprised if they wanted to milk every last drop of telemetry that the public release brings before committing to any decisions on V2.
Let's also not forget that the front visors need to connect magnetically to the center camera console attached to the windshield. Changing the size of this camera box would require a redesign of the front visors, something they'd likely not want to do. For them to launch the Founder's edition with a visor that's a different size seems unnecessarily complicated. So, unlike with the S, a retrofit to a larger dual camera box would also require new visors.
- - - Updated - - -
Agreed. I think the big difference is speculation verses what's actually in the cars right now. Things change during a long, complicated development. If the intention was there but costs or marketing data changed the direction, you can defend that. However, being incorrect about the hardware in the actual cars today is another thing. And why even comment if the intention is to surprise people with new hardware soon? He could've easily ignored the tweet... but specifically chose to clarify that the hardware and capabilities are the "same."
If you look at a close-up of the X models that were already delivered, you can see that the camera housing is already big enough to hold two cameras. No new visors would be needed.
Picture from the X reveal:
Picture of the X from Tesla's website showing two cameras:
I would guess that Telsa has no reason to deploy V2 of autopilot for the next couple of years. What I mean is that if V1 Autopilot works as well as I think it will... allowing the car to 'drive itself' when on highways and maybe sometimes within a city (for short periods of time), V2 would only go to waste. The law/insurance aspects may step in if a car truly starts driving itself. Sure, Tesla could add in the 2nd camera (or whatever V2 consists of) right now, but in ~2 years time, when they finally start to release V2 software, I am sure the technology available for the hardware will be leaps improved from what we have now. Why would Tesla want to invest the capital on hardware that will go unused for years, and won't be as good as 'the competition' come the time it matters? Yes, I realize they somewhat did that with V1 Autopilot... some people with mention it coming out ~1 after the hardware was available. Perhaps the 'final' AP was a year late, yes, but we've been using the sensors for most of life of the car (TACC, auto high beams, blind spot).
Until we experience AP V1, it is hard to really gauge what it will be. TACC took some people a while to get accustomed to. Autopilot will certainly challenge some peoples habits. It could also need people to develop new habits. With TACC, it is typically very obvious that the car won't stop at a traffic light/stop sign (unless a car is in front of you). Think of a person driving an Autopilot equipped Tesla for the first time, coming from a car with just normal cruise. They turn on AP to test it out, slowly build trust that it is keeping them in their lane under most road conditions. Sure, their hands may need to be on the wheel, but if the wheel is moving on its own, you just go with it. Then as you come up to the first red light with a car in front of you and the car stops by itself (and then resumes by itself when traffic moves again). I could foresee a situation where after the 2nd or 3rd red light (with a car stopped in front of you) where your car handles (nearly) everything for you, you could just expect it to stop for you at the next red light (but with no car in front to cause the stopping). I hope there are no accidents like this, where Autopilot gets the blame and puts the hold system under the microscope and perhaps freezes progress.
Aside from habits that may need to change/built, surely AP V1 will also keep improving. Even if it isn't intended to 'drive itself', trying to figure out how to handle the multitude of variables it is going to face will likely keep AP V1 evolving for the next year or two or more. If Tesla can keep busy improving the core logic of AP for the foreseeable future, again I reckon it doesn't make sense for them to put in V2 hardware anytime soon. At some point they may feel like they have brought AP V1 as far as they can, and then V2 HW makes sense (even if silently installed... ie not there to dramatically improve Autopilot, but just act as additional sensors at first, while giving beta testers the ability to start on feedback for V2 AP/autonomous driving).
I don't know about any of that. I'm just saying that the photo on Tesla's web site shows two cameras, the Founders X vehicles have room for a 2nd camera, and it's logical to deduce that it's most likely coming. That does not take away anything from the vehicle as it currently stands, and makes no statement about the desirability of Model S or X. Tesla held back a lot of things from the Model S that were in development, if you want to look at things that way... let's see... heated steering wheel, parking sensors, Autopilot sensors, AWD. Oh wait, Tesla didn't hold those things back. They were incremental improvements. The same thing will happen here. For whatever reason, Tesla is not ready to ship the car with two cameras. However, the space is there and Tesla is clearly planning for it.
Some would disagree if you replace "camera" with "motor".
These are important, but Tesla is a multifaceted company and surely has separate teams working on their own goals.
I would argue that in order to remain relevant in the automotive space, they're going to have to continue to push the abilities of autopilot. Tesla has an amazing opportunity right now, which isn't often discussed. They have much more than an edge at this point over other manufacturers when it comes to AutoPilot and the goal of autonomous driving. What's the edge? Deployment and data collection. Here's how that looks to me:
- Google has a lot of miles clocked and a lot of research put into the field. However, they don't have a fleet the size of the MS autopilot enabled fleet, which means they are limited in the overall size or quality of the supervised learning model they can build.
- In the same way Captcha has every one of us doing tiny bits of work on OCR supervised learning projects, Tesla has every one of those cars with the ability to feed back supervised learning data to the mothership. Obviously I don't know how this is done, but if it were me, it would include GPS coordinates, steering feedback, and some feature set of the camera. As MS drivers cover lots of ground, the model continually is trained in what people do in those locations. With machine learning, most of the time more data means a more accurate model that is more generalizable to other scenarios.
- Other manufacturers can get on this quickly as their vehicles begin being outfitted with always-on internet connections. They will have the ability to rapidly overtake Tesla's models due to the production numbers.
I feel that Tesla has a lot to gain by adding as many sensors as possible in order to better build and train autopilot and autonomous driving models.
Agreed. I'd disagree too because the front and rear motors are different. Two side by side cameras would use the same hardware, but not the same software.