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

What's the likelihood we will see Speed Limit Sign recognition?

This site may earn commission on affiliate links.
Then how do you explain AP 2.0+ not reading speed limit signs? They obviously have the technical capability to do this... AND this would be immensely useful for customers.
AP2 is capable of reading signs, but, as referenced in some of the earlier posts, Mobileye managed to get a patent on reading signs. AP1 Teslas use Mobileye, so they may read signs. AP2 and beyond is Tesla's own software and so would have to pay a royalty to Mobileye (if they would even agree to license to Tesla) in order to read signs.
Other car makers can read signs because they use Mobileye software.
 
AP2 is capable of reading signs, but, as referenced in some of the earlier posts, Mobileye managed to get a patent on reading signs. AP1 Teslas use Mobileye, so they may read signs. AP2 and beyond is Tesla's own software and so would have to pay a royalty to Mobileye (if they would even agree to license to Tesla) in order to read signs.
Other car makers can read signs because they use Mobileye software.

The only caveat is that Intel now owns MobileEye along with all their patents.

I can't imagine Intel being nasty towards Tesla, despite the complication relationship between Tesla/MobileEye.

Besides I bet Tesla now has a bunch of FSD related patents of their own by now.
 
The challenge is not only in reading the speed limit signs - but also in determining which signs actually apply to your lane/road.

On our local freeways, there are HOV lanes, main lanes, exit ramps and frontage roads - all adjacent with little spacing between them. When a speed limit sign is placed between these lanes, it's not always obvious to which lanes the speed limit applies.

The lack of standardization in speed limit signs and placement makes this even more challenging.

To workaround the Mobileye patent, Tesla has tried to use an offline speed limit database coupled with the average speed of Tesla vehicles on specific roads. Neither of these data sources are 100% reliable. When Tesla changed speed limit database providers, many roads had missing or incorrect speed limits - and that problem continues (especially in areas with current or recent construction). The fleet speed doesn't reflect the speed expected under the current conditions - which can change based on traffic, weather, ...

The Mobileye patent appears to be an example of a patent granted for an obvious feature (like swiping your finger on a smartphone screen), which unfortunately makes it difficult or impossible to implement a solution without using Mobileye's software or paying them royalties.

Tesla is running out of time to get this addressed - as they move closer to seeking FSD approval (at least in limited use), because it's unlikely any FSD system will be approved unless the manufacturer can guarantee the vehicle will obey traffic laws, including posted speed.

I suspect Tesla actually has this feature working - and Musk has been driving with it for a while - but without resolving the Mobileye patent, they can't deploy this feature for AP2/AP3 vehicles...
 
It’s not their data. I’m sure they just get it from one of the other bit mapping companies.
No reason Tesla couldn't keep their own database - I think they actually have a fork of OSM data that they are merging in all sorts of other sources into. Or push it back to the company they are getting the maps from. That sort of data would have lots of value.
 
Maybe Tesla could go around this patent by reading speed limit
Enforceable or not, Tesla doesn’t have the resources or bandwidth right now to take on Intel to find out. That would be suicide.
So you Tesla plans can have FSD without sign recognition, or do you they will just forget auto-pilot in order not to take on Intel (which according to you would be suicide)?
 
So you Tesla plans can have FSD without sign recognition, or do you they will just forget auto-pilot in order not to take on Intel (which according to you would be suicide)?

“Right now” is an important part of my statement. Read it again before you pigeonhole me with silly absolutes.

Eventually they’ll have to do something about it. Suck it up and pay to license, pick a fight, code around it, or maybe some combination of the three. But right now it’s just a distraction given how far off the rest of the supporting elements are. Put another way, fixing the lack of speed limit sign recognition is NOT the critical path to FSD for the foreseeable future.
 
“Right now” is an important part of my statement. Read it again before you pigeonhole me with silly absolutes.

Eventually they’ll have to do something about it. Suck it up and pay to license, pick a fight, code around it, or maybe some combination of the three. But right now it’s just a distraction given how far off the rest of the supporting elements are. Put another way, fixing the lack of speed limit sign recognition is NOT the critical path to FSD for the foreseeable future.
According to Elon robotaxis are coming next year, which gives them no more than 13 months to finish the fight. Legal fights usually take way longer than that, so if they want to use it in 13 months, they should already be fighting it. Of course, if you believe Elon was lying or smoking something potent when he committed to robo-taxis by the end of 2020, then sure, no reason to fight it now, let Waymo (Google) fight it out in the next few years, hopefully clearing the way by 2030, 2040 or whenever FSD will actually come for AP2, AP2.5 and AP3 cars.
 
“Right now” is an important part of my statement. Read it again before you pigeonhole me with silly absolutes.

Eventually they’ll have to do something about it. Suck it up and pay to license, pick a fight, code around it, or maybe some combination of the three. But right now it’s just a distraction given how far off the rest of the supporting elements are. Put another way, fixing the lack of speed limit sign recognition is NOT the critical path to FSD for the foreseeable future.
FSD is feature complete next month! :p They better figure it out.
 

Interesting. Each of the independent claims states that the image processor must both detect the traffic sign and perform another driver assistance function.

If Tesla dedicated one "processor" on the FSD computer to signs only (perhaps with other stuff not related to driver assistance) and the other processor to driving assistance, they might not infringe. I guess it depends on what the meaning of the word "processor" is. ;)