Welcome to Tesla Motors Club
Discuss Tesla's Model S, Model 3, Model X, Model Y, Cybertruck, Roadster and More.
Register

Autopilot 2.0 Not Imminent Based On Production Model X Design Studio [Speculation]

This site may earn commission on affiliate links.
My point is that they'll obviously be building a number of test units (likely in the hundreds or thousands) over the course of the next few years, but that doesn't mean they have to release the hardware before getting regulatory approval... They could just as easily demonstrate the tech/functionality without releasing anything to the public. That way they can hopefully release the hardware and software TOGETHER.

You are driving the test units.

any hardware will get rolled into the fleet quietly and software will run in the background collecting data.

you are ignoring the past history of Tesla and how they will obviously do the next iteration.
 
Yep, I know. But I mean fully functional automatic emergency braking from HIGHWAY speeds. Just like one in new Volvo XC90. So it could maybe prevent a crash from higher speeds and not only city speeds.

- - - Updated - - -

Model S has automatic emergency braking. Single camera and radar


I was suppoused to quote this. Just wondering why that highway is mentioned. Never seen that in Model S. Could there be a better hardware in X for that.
 
I was suppoused to quote this. Just wondering why that highway is mentioned. Never seen that in Model S. Could there be a better hardware in X for that.

From the manual [page 80], so yes, Model S has AEB at highway speeds.

"When Automatic Emergency Braking has reduced the driving speed by 25 mph (40 km/h), the brakes are released. For example, if Automatic Emergency Braking applies braking when driving at 56 mph (90 km/h), it releases the brakes when the speed has been reduced to 31 mph (50 km/h). Automatic Emergency Braking operates only when driving between 5 mph (8 km/h) and 85 mph (140 km/h)."
 
I agree that they don't HAVE to release the hardware before regulatory approval, but if they do (like they did with AP), it gives them tens of thousands of cars equipped with Level 4 sensors, collecting and streaming data back to the mothership. It's much "cheaper" than building out a thousand prototypes, hiring 1,000 employees to drive millions of miles across the world.

My point is to demonstrate it works 100% of the time, which is what Level 4 is supposed to do, you need a lot of data. To get that lot of data, the cheapest/easier solution is to do what they did with AP and release the hardware without telling anyone, and collect data (they might be doing some of this now), and test/build your algorithms on that.

I see your point, it could happen your way and time will tell, but IMHO hardware will come very soon, and in X years we'll get the software, insuring a lot of free data collection for Tesla.

I doubt they are going to upgrade the hardware much before the software is ready to use it. I know they did for 7.0, but remember that they thought that 7.0 was only a few months away (even though it turned out to be a year). The software for sure is a few years away, and I expect the major hardware rev to be nearly that as well. (It's what I'm banking on anyway.) Might see newer-gen versions of what we have, but nothing too major before then.

There are profit issues that are impacted by throwing more expensive hardware in sooner than it can be used, and also codebase issues if you have to support current-gen AP in two different sensor arrays.

Still, if the costs aren't too out of whack and they can keep the other issues in line, I could see them wanting to field test the hardware. (But if they're field testing it, it implies that it might need to change over time, and then the whole benefit of releasing it early is gone or worse).
 
Last edited:
Model S and X have AEB at speeds from 5 to 85mph. I am unclear on why you think they do not have AEB at highway speeds.
On that topic, I'm curious as to how it functions.... Does it use the camera/sensors, one or the other? Also, how much stopping time does it really have at 85MPH? Not much... I fail to see how it's functional at high speeds.
 
As promised. Red = AP camera, same as Model S (it has a reflection over it, but you can see it). Orange = empty space for 2nd camera, but it's not there.

View attachment 105134

Thanks for that. I don't get why Tesla would want to risk falling behind other automakers by only using one forward facing camera when it's cheap to install another. I guess on the other hand they may not have wanted to rock the boat again after what happened with autopilot 1.0. I wonder if there's a wiring harness behind there for the other camera? Then it could be like the LTE upgrade at the service centers.

I thought the webpage ordering picture shows an actual camera in that spot? Is that correct?
 
Thanks for that. I don't get why Tesla would want to risk falling behind other automakers by only using one forward facing camera when it's cheap to install another. I guess on the other hand they may not have wanted to rock the boat again after what happened with autopilot 1.0. I wonder if there's a wiring harness behind there for the other camera? Then it could be like the LTE upgrade at the service centers.

I thought the webpage ordering picture shows an actual camera in that spot? Is that correct?

I don't get it either. And the Model X webpage still shows two cameras...
 
Thanks for that. I don't get why Tesla would want to risk falling behind other automakers by only using one forward facing camera when it's cheap to install another. I guess on the other hand they may not have wanted to rock the boat again after what happened with autopilot 1.0. I wonder if there's a wiring harness behind there for the other camera? Then it could be like the LTE upgrade at the service centers.

I thought the webpage ordering picture shows an actual camera in that spot? Is that correct?

So far, all of the dual front camera systems I've seen are stereo pairs - two identical cameras with matching lenses, set a distance apart.

Stereo cameras are used to determine the distance to objects by analyzing exactly where in each camera's image the object appears - things at infinite distance show up in the exact same place, while things that are closer are offset progressively toward the "center" sides of the images (assuming the cameras are boresighted exactly - this isn't actually necessary as long as they are calibrated to understand what's the same spot on each one.)

Implicit in this is that the accuracy of the distance measurement is greatly influenced by the distance between the cameras (along with the camera resolution and the accuracy of placement/calibration.) All of the stereo sets I've seen are several inches apart (some even a couple of feet) - but the "second camera" location on the X is only an inch or two away from the existing camera - which makes stereo cameras less practical.

Most of the time, the stereo cameras are an alternative to radar for adaptive cruise type purposes - the only exception I'm aware of being the Mercedes system that uses the stereo cameras to look for road surface contour changes for the suspension system. Tesla seems happy with their radar installation, which is typical of the class/price range.

I tend to think what Tesla really needs for Autopilot 2.0 is more coverage around the car, to look sideways at lights/intersections, to look far behind for lane changes. Some of the side looking could be achieved by a second front camera with a much wider angle, but really it should be dedicated side-looking camera I would think. The rear view could be either a second camera with a much longer lens or radar/lidar.
Walter
 
So far, all of the dual front camera systems I've seen are stereo pairs - two identical cameras with matching lenses, set a distance apart.

Stereo cameras are used to determine the distance to objects by analyzing exactly where in each camera's image the object appears - things at infinite distance show up in the exact same place, while things that are closer are offset progressively toward the "center" sides of the images (assuming the cameras are boresighted exactly - this isn't actually necessary as long as they are calibrated to understand what's the same spot on each one.)

Implicit in this is that the accuracy of the distance measurement is greatly influenced by the distance between the cameras (along with the camera resolution and the accuracy of placement/calibration.) All of the stereo sets I've seen are several inches apart (some even a couple of feet) - but the "second camera" location on the X is only an inch or two away from the existing camera - which makes stereo cameras less practical.

Most of the time, the stereo cameras are an alternative to radar for adaptive cruise type purposes - the only exception I'm aware of being the Mercedes system that uses the stereo cameras to look for road surface contour changes for the suspension system. Tesla seems happy with their radar installation, which is typical of the class/price range.

I tend to think what Tesla really needs for Autopilot 2.0 is more coverage around the car, to look sideways at lights/intersections, to look far behind for lane changes. Some of the side looking could be achieved by a second front camera with a much wider angle, but really it should be dedicated side-looking camera I would think. The rear view could be either a second camera with a much longer lens or radar/lidar.
Walter

Right but if redundancy is the issue (perhaps a legislative requirement for intersections) then the location of a second camera becomes less important.
 
Right but if redundancy is the issue (perhaps a legislative requirement for intersections) then the location of a second camera becomes less important.

I haven't seen any indication that redundancy is a concern - and the camera should be at least as reliable as any of the other system elements - radar, ultrasound, computers.

In an aerospace application you'd have a three node "I tell you three times" voting computer group and dual sets of all the inputs and outputs, but in an automotive application I suspect that all they'll need are systems that can correctly detect and identify failures and pull over safely.

All of the speculation I'd read about AP 2.0 was about adding capability rather than meeting regulatory requirements (which are currently mostly undefined, AFAIK.)
Walter
 
I haven't seen any indication that redundancy is a concern...

"Here’s Musk’s full response, when asked if it would be possible to move toward greater levels of autonomy in the Tesla Model S without adding hardware:

There’s more that can be done, but the sensor suite is not the full autonomy suite. For full autonomy you’d obviously need 360 cameras, you’d probably need redundant forward cameras, you’d need redundant computer hardware, and like redundant motors and steering rack. For full autonomy you’d really want to have a more comprehensive sensor suite and computer systems that are fail proof."


(My emphasis.)

The article goes on to read (with more of my emphasis):

"A future Tesla car would need 360 degree cameras, according to Musk, as well as redundant forward cameras, redundant computer systems, and a redundant motor to act as a failsafe. As Musk has noted many times in the past, full autonomy will require the approval of many regulatory bodies,..."

Of course, no one knows what regulatory approval will require, but redundancy of dual forward cameras seems to be a pretty good guess.

From:

Elon Musk says that the LIDAR Google uses in its self-driving car | 9to5Google
 
Last edited:
"Here’s Musk’s full response, when asked if it would be possible to move toward greater levels of autonomy in the Tesla Model S without adding hardware:

There’s more that can be done, but the sensor suite is not the full autonomy suite. For full autonomy you’d obviously need 360 cameras, you’d probably need redundant forward cameras, you’d need redundant computer hardware, and like redundant motors and steering rack. For full autonomy you’d really want to have a more comprehensive sensor suite and computer systems that are fail proof."


(My emphasis.)

From:

Elon Musk says that the LIDAR Google uses in its self-driving car | 9to5Google

And since I hadn't seen that, my disclaimer was true then. Thanks for the link. :)

I still tend to think the 360 cameras would be the first step, to expand capability while still in semi-autonomous operation.

No car currently in operation has redundant steering equipment or fail-proof computers for any aspect of operation that I'm aware of, and the only car with any sort of redundant power train is the dual motor Model S/X - and even then it is only for certain failures of the front motor only.

I think Elon may be setting an excessive standard considering the state of the industry - but if he can manage it at a reasonable price I won't complain.
Walter