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

Ford CEO promises L3 highway by 2026

This site may earn commission on affiliate links.

diplomat33

Average guy who loves autonomous vehicles
Aug 3, 2017
12,849
18,942
USA
Ford CEO says Blue Cruise will be L3, or eyes-off, hands-off, on highways, up to 80 mph, only in good weather, by 2026. Sounds like an Elon FSD promise. LOL.

“We’re getting really close,” Farley said in a May 31 interview with Bloomberg TV’s David Westin. “We can do it now pretty regularly with a prototype, but doing it in a cost-effective way is just the progress we’re going to need to make.”

“Level 3 autonomy will allow you to go hands and eyes off the road on the highway in a couple years so then your car becomes like an office,” Farley said. “You could do a conference call and all sorts of stuff.”

Farley suggested Ford’s system would operate at speeds of up to 80 miles per hour on the highway, but only under clear skies.

“We only think we can do it on sunny days,” Farley said. “Heavy rain and stuff makes it difficult to do it at 80 miles an hour.”

Source: Bloomberg Article


Ford_Mustang_Mach-E_AWD_First_Edition_–_h_15072021 (1).jpg

"Ford Mustang Mach-E AWD First Edition – h 15072021" by © M 93 is licensed under CC BY-SA 3.0.
Admin note: Image added for Blog Feed thumbnail
 
It might be a strategy to not answer the present issues by boasting future non-existent competency.


It's probably good PR. This way, Farley changes the conversation and gets people talking about this amazing future product instead of the present issues.

Having said that, if Ford really has a L3 prototype and they spend the next 2 years, improving and validating safety, it is not inconceivable that they deliver some L3, maybe not in 2 years, maybe in 3. After all, AV progress is pretty rapid now. Engineers can do a lot in 2-3 years.
 
Yes, sounds very much like an Elon promise, with one key exception: he's setting a very achievable mid-term milestone and not trying to boil the ocean all at once by promising fully autonomous robotaxis in a year or two. And even more to the point, it's an actually useful feature.

My vision for autonomous travel is two part:
  1. In the nearer term, I am looking for on-ramp to off-ramp autonomy. The use case here is a long trip, mostly on highway, where I don't want to be bothered by the extremely mundane driving tasks (even better, I'd like to be able to do this trip overnight while I sleep). It's perfectly fine with me that it may be limited to limited access highways (or even just interstates), as long as it is reliable and I don't have to wait 20 years for the technology to reach a point that it is truly hands off. This seems very achievable in the near- to mid-term.
  2. By the time I am too old to drive myself, I would like a fully autonomous vehicle to drive me to my doctor's appointments or even run errands for me (I envision a future grocery store trip will consist of me placing an order online and then sending my car to retrieve the order, with a person at the store loading up my vehicle with my order). I (hopefully) have plenty of time for this to become a reality.
This seems like a reasonable progression that gives me some desirable functionality in the nearer term.

Elon seems to be focusing strictly on #2, hoping that #1 comes along for the ride. He is continually mistaken that #2 is right around the corner (if it truly was, there might be some logic to this), and in fact this focus has made the #1 experience in a Tesla LESS attractive in recent months.
 
Yes, sounds very much like an Elon promise, with one key exception: he's setting a very achievable mid-term milestone and not trying to boil the ocean all at once by promising fully autonomous robotaxis in a year or two. And even more to the point, it's an actually useful feature.

My vision for autonomous travel is two part:
  1. In the nearer term, I am looking for on-ramp to off-ramp autonomy. The use case here is a long trip, mostly on highway, where I don't want to be bothered by the extremely mundane driving tasks (even better, I'd like to be able to do this trip overnight while I sleep). It's perfectly fine with me that it may be limited to limited access highways (or even just interstates), as long as it is reliable and I don't have to wait 20 years for the technology to reach a point that it is truly hands off. This seems very achievable in the near- to mid-term.
  2. By the time I am too old to drive myself, I would like a fully autonomous vehicle to drive me to my doctor's appointments or even run errands for me (I envision a future grocery store trip will consist of me placing an order online and then sending my car to retrieve the order, with a person at the store loading up my vehicle with my order). I (hopefully) have plenty of time for this to become a reality.
This seems like a reasonable progression that gives me some desirable functionality in the nearer term.

Elon seems to be focusing strictly on #2, hoping that #1 comes along for the ride. He is continually mistaken that #2 is right around the corner (if it truly was, there might be some logic to this), and in fact this focus has made the #1 experience in a Tesla LESS attractive in recent months.

Level 3 doesn't allow you to sleep. That's Level 4. Realistically, I think you're not going to get Level 4 restricted highway. The problem is that you would not always be safe to drive if you are forced to divert off the highway.
 
Last edited:
My vision for autonomous travel is two part:
  1. In the nearer term, I am looking for on-ramp to off-ramp autonomy. The use case here is a long trip, mostly on highway, where I don't want to be bothered by the extremely mundane driving tasks (even better, I'd like to be able to do this trip overnight while I sleep). It's perfectly fine with me that it may be limited to limited access highways (or even just interstates), as long as it is reliable and I don't have to wait 20 years for the technology to reach a point that it is truly hands off. This seems very achievable in the near- to mid-term.
  2. By the time I am too old to drive myself, I would like a fully autonomous vehicle to drive me to my doctor's appointments or even run errands for me (I envision a future grocery store trip will consist of me placing an order online and then sending my car to retrieve the order, with a person at the store loading up my vehicle with my order). I (hopefully) have plenty of time for this to become a reality.
This seems like a reasonable progression that gives me some desirable functionality in the nearer term.
+1. My hopes/expectations as well.
 
Yes, sounds very much like an Elon promise, with one key exception: he's setting a very achievable mid-term milestone and not trying to boil the ocean all at once by promising fully autonomous robotaxis in a year or two. And even more to the point, it's an actually useful feature.

My vision for autonomous travel is two part:
  1. In the nearer term, I am looking for on-ramp to off-ramp autonomy. The use case here is a long trip, mostly on highway, where I don't want to be bothered by the extremely mundane driving tasks (even better, I'd like to be able to do this trip overnight while I sleep). It's perfectly fine with me that it may be limited to limited access highways (or even just interstates), as long as it is reliable and I don't have to wait 20 years for the technology to reach a point that it is truly hands off. This seems very achievable in the near- to mid-term.
  2. By the time I am too old to drive myself, I would like a fully autonomous vehicle to drive me to my doctor's appointments or even run errands for me (I envision a future grocery store trip will consist of me placing an order online and then sending my car to retrieve the order, with a person at the store loading up my vehicle with my order). I (hopefully) have plenty of time for this to become a reality.
This seems like a reasonable progression that gives me some desirable functionality in the nearer term.

Elon seems to be focusing strictly on #2, hoping that #1 comes along for the ride. He is continually mistaken that #2 is right around the corner (if it truly was, there might be some logic to this), and in fact this focus has made the #1 experience in a Tesla LESS attractive in recent months.
Yeah, that's what I thought 20 years ago. Then I got one. I no longer have any expectations of self driving within my lifetime. Oddly enough, the car will certainly outlast me. It doesn't get driven very much and seems to be mechanically simple and very little to break. The service manual seems adequate to use to fix anything that might break, (which is good, since the service 50 miles away just responds with, "they all do that" and closes the service request ticket). Tesla car insurance is now the largest single expense in my life and slightly more than my health insurance.
 
Elon says he ain’t let’n no Jr Jim Blue Cruise 💩 on his BS degree in hyperbole.

View attachment 1053880
Sounds pretty reasonable. If it takes me a over a year to get mine out of the carport. I usually don't get interventions inside the carport, (if you don't count trying to kill the dry windshield wipes). Last August I made it clear out of the driveway once without an intervention.
 
  • Funny
Reactions: Dracaris
..."starting to get to the point where once known bugs are fixed"...
"If we had ham, we could have ham and eggs, if we had eggs!"

The thing about L3 is that it requires just as high a degree of reliability as L4, just at a different task. L4 must be able to make correct driving decisions (or in extreme situations, go into safe mode and park itself) 99.99999% of the time. L3 must be correctly able to identify when it's about to exceed its ODD (so it knows when to hand back control) 99.99999% of the time. Tesla may currently be at 99% (per mile, say) for both of these tasks, but that still leaves a LOT of 9's before it's even L3-ready.

Also, as a software engineer who's worked on lots of complicated projects, it's a truism that no project EVER ships with all or even most of the serious known bugs fixed. Just because Tesla has identified a particular bug doesn't mean they'll be able to solve it. This is particularly so for neural networks, because there's no line of code you can point to and say "Aha, THERE'S the bug!" It's a lot more squishy, and improving one behavior or edge case can often cause regressions in others.

It's not clear what hardware suite these Blue Cruise L3 cars will have; Farley said their main hurdle is cost-effectiveness, which probably implies Lidar and such. It's plausible that they could have a narrow-ODD L3 (Lidar-mapped highways, good weather only) with an overbuilt-sensor-suite car, by the end of 2026. Could Tesla do the same with HW4? I think if they were to redirect their entire focus on solving the same narrow L3 ODD, then maybe, but Elon seems exclusively focused on wide-ODD L4, which is quite a ways further off. I'll be very pleasantly surprised if Tesla publicly deploys a useful-ODD L4 system with HW6 by 2030.
 
"If we had ham, we could have ham and eggs, if we had eggs!"

The thing about L3 is that it requires just as high a degree of reliability as L4, just at a different task. L4 must be able to make correct driving decisions (or in extreme situations, go into safe mode and park itself) 99.99999% of the time. L3 must be correctly able to identify when it's about to exceed its ODD (so it knows when to hand back control) 99.99999% of the time. Tesla may currently be at 99% (per mile, say) for both of these tasks, but that still leaves a LOT of 9's before it's even L3-ready.

Also, as a software engineer who's worked on lots of complicated projects, it's a truism that no project EVER ships with all or even most of the serious known bugs fixed. Just because Tesla has identified a particular bug doesn't mean they'll be able to solve it. This is particularly so for neural networks, because there's no line of code you can point to and say "Aha, THERE'S the bug!" It's a lot more squishy, and improving one behavior or edge case can often cause regressions in others.

It's not clear what hardware suite these Blue Cruise L3 cars will have; Farley said their main hurdle is cost-effectiveness, which probably implies Lidar and such. It's plausible that they could have a narrow-ODD L3 (Lidar-mapped highways, good weather only) with an overbuilt-sensor-suite car, by the end of 2026. Could Tesla do the same with HW4? I think if they were to redirect their entire focus on solving the same narrow L3 ODD, then maybe, but Elon seems exclusively focused on wide-ODD L4, which is quite a ways further off. I'll be very pleasantly surprised if Tesla publicly deploys a useful-ODD L4 system with HW6 by 2030.
What's the human number?
It's clearly over 99%. But how many more 9's is the best human? 2,3,4 maybe?
So the real number has to use humans as the base. Otherwise, how can you accurately pick a goal number?

The day the overall system is even just equal to the best humans, then it's ready for deployment.
Because then, every driver on the road is worse than the system. And if it's better than any human at making the right split-second decision in X% of situations, then it will save lives. And the system will only continue to improve over time.
So why wait longer?

If the best human driver is at 99.999% accurate. Then that's the goal for autonomous driving systems (ADS).
If the human number is 99.9%, then that's the ADS goal.
Because at whatever that level is, it's already better than any human. Why would it need to be X% better than humans?
 
Last edited:
What's the human number?
An interesting question is whether the bar is human driven or human+machine driven. It seems possible that human driving with machine assist is unbeatable.
 
So Im shopping for a Sprinter or a Transit. FSD is a functionality is important. I decided to put off the purchase until 2025 but now maybe I can get by for another 2 years. To me this all marketing. If Tesla can’t do it yet with all the resources they have committed to what are the chances Ford can? Have they committed as much as Tesla??
 
An interesting question is whether the bar is human driven or human+machine driven. It seems possible that human driving with machine assist is unbeatable.
If that's the case, then why develop past L3?
The point was for RT's and if we don't want to have to pay attention.
Also, I don't believe that eventually keeping human involvement will be a positive thing. For now, maybe, but eventually, no. The system will be too fast and able to determine the best possible outcome instantly, where a human's input will just get in the way.
 
If that's the case, then why develop past L3?
The point was for RT's and if we don't want to have to pay attention.
Also, I don't believe that eventually keeping human involvement will be a positive thing. For now, maybe, but eventually, no. The system will be too fast and able to determine the best possible outcome instantly, where a human's input will just get in the way.
You don’t have to pay attention with L3.
I doubt many collisions are primarily caused by slow reaction time. I think people really want self-driving cars and don’t want machine nannies so we won’t set the bar too high.
 
  • Like
Reactions: APotatoGod
So Im shopping for a Sprinter or a Transit. FSD is a functionality is important. I decided to put off the purchase until 2025 but now maybe I can get by for another 2 years. To me this all marketing. If Tesla can’t do it yet with all the resources they have committed to what are the chances Ford can? Have they committed as much as Tesla??
Yes, Ford certainly can. The reason is that Ford is not bound by Elon's wannabe-engineer limitations on the sensor suite. Elon has crippled FSD by mandating vision only. But... but.... People can drive with only their eyes so cars should be able to as well! While this makes intuitive sense, it just doesn't work in the real world. Tesla can't even figure out TACC using only vision. I still get multiple phantom braking episodes per drive. FSD is a pipe dream with only vision. The other makers that are open to using a mix of the best sensors will leapfrog past Tesla.