What SAE Level 3-5 system will be released to consumers in 2020?

[Daniel in SD]

I’m a skeptic. I don’t think anyone will release an eyes off the road system to consumer vehicles in 2020.
https://www.daimler.com/innovation/case/autonomous/drive-pilot-2.html
Audi Newsroom

 

[diplomat33]

I’m a skeptic. I don’t think anyone will release an eyes off the road system to consumer vehicles in 2020.
DRIVE PILOT | Daimler
Audi Newsroom

I would not be surprised if Audi or another auto maker delivers some type of L3 or L4 autonomous driving system by 2020 or 2021. It should be noted that these systems have limited ODD, as they only work on limited access highways.

Here is a video of a test drive from 2 years ago, of "Jack", Audi's L4 highway system that is mentioned in your link:

It appears that Audi had a hands off highway driving prototype 2 years ago. So I would not be surprised if they deliver by 2020 of 2021.

I want to be optimistic about Tesla. Since I own a Tesla, I would love it if Tesla achieved L3 ot L4 highway by 2020. I do think that if Tesla put "city NOA" on pause and focused entirely on highway driving, they could probably get NOA up to L3 highway by end of 2020. But since they are focused on a more general solution that is not limited to just highway driving, I feel like Tesla may get a system that has a bigger ODD (can be used on both highways and city streets) but does not quite crack the L3 barrier by 2020 and thus would still require hands on wheel.

I definitely think 2020 will be an interesting year for autonomous driving. It is exciting to see the progress by so many different companies.

 

[DanCar]

Audi Newsroom
The Audi article is laughable. Level 3 in 2017? Typos: enadling

 

[diplomat33]

Audi Newsroom
The Audi article is laughable. Level 3 in 2017?

I don't think that is a mistake. Audi's Traffic Jam Pilot is technically L3. But it has a very small ODD to make it L3. It only works under very narrow conditions:

"In 2017, Audi will introduce what’s expected to the world’s first to-market Level 3 automated driving system with “Traffic Jam Pilot” in the next generation Audi A8. The system will give drivers the option to travel hands-free up to 35 mph, when certain conditions are met prior to enadling this feature — for instance, the vehicle will ensure it is on a limited-access, divided highway."

 

[DanCar]

I don't think that is a mistake. Audi's Traffic Jam Pilot is technically L3. But it has a very small ODD to make it L3. It only works under very narrow conditions:

"In 2017, Audi will introduce what’s expected to the world’s first to-market Level 3 automated driving system with “Traffic Jam Pilot” in the next generation Audi A8. The system will give drivers the option to travel hands-free up to 35 mph, when certain conditions are met prior to enadling this feature — for instance, the vehicle will ensure it is on a limited-access, divided highway."

That definitely didn't make it to the U.S.

 

[Daniel in SD]

I don't think that is a mistake. Audi's Traffic Jam Pilot is technically L3. But it has a very small ODD to make it L3. It only works under very narrow conditions:

"In 2017, Audi will introduce what’s expected to the world’s first to-market Level 3 automated driving system with “Traffic Jam Pilot” in the next generation Audi A8. The system will give drivers the option to travel hands-free up to 35 mph, when certain conditions are met prior to enadling this feature — for instance, the vehicle will ensure it is on a limited-access, divided highway."

I think that many Tesla owners in California would love a system that did that (we have a lot of slow freeways here!). Unfortunately Audi still hasn't released it (even in Germany).
The Audi A8 has LIDAR so they were probably thinking that a simple "don't run into anything" algorithm will work at 35mph on limited-access roads. I bet they ran in to more corner cases than expected.

 

[Daniel in SD]

That definitely didn't make it to the U.S.

It didn't make it anywhere.

 

[DanCar]

....DRIVE PILOT | Daimler ...

The Merc article doesn't give a timeline, just says they are working on it like everyone else. The Audi article is stale. Tesla doesn't give a timeline either. There is no reason to be optimistic that Level 3 will come soon. I'll say the chances of Merc or Audi releasing Level 3 in the U.S. in the next two years is close to zero.
Nissan says they will have it available next year: Toward fully autonomous drive
Mobileye says Level 3 in 2018, available for all automakers: The Evolution of EyeQ - Mobileye And level 4-5 coming next year. Another example of a stale article.

 

[Daniel in SD]

The Merc article doesn't give a timeline, just says they are working on it like everyone else. The Audi article is stale. Tesla doesn't give a timeline either. There is no reason to be optimistic that Level 3 will come soon. I'll say the chances of Merc or Audi releasing Level 3 in the U.S. in the next two years is close to zero.
Nissan says they will have it available next year: Toward fully autonomous drive
Mobileye says Level 3 in 2018, available for all automakers: The Evolution of EyeQ - Mobileye And level 4-5 coming next year. Another example of a stale article.

Drive Pilot was supposed to be a feature of the 2020 S Class but it doesn't look like there's any news about it since it was announced late 2018.
Tesla has said they will have robotaxis by the end of 2020 but I suppose it's possible those will not be sold to consumers .
I think if a German car company releases a system in Germany it will come to the US pretty shortly thereafter (assuming it's much safer than a human of course).

 

[Saghost]

I think FSD will come to an inflection point in 2020.

Tesla has been building to this point for over three years, ever since they started really scoping the problem. They now have the computers they thought they'd need and a fleet of cars equipped with them to learn on.

So either they judged the problem correctly, and FSD will show up in 2020, or the Tesla experts misjudged the scope of the problem, and they won't have FSD for at least a few years, until they release a new set of computers and sensors (in which case there's a good chance someone else will solve the problems first.)

I'm not 100% sure which just yet.

 

[DanCar]

... or the Tesla experts ...

It is only Elon dreaming.

... FSD will show up in 2020 ...

Elon changed the goal post again after making that statement. His latest one is three years.

 

[diplomat33]

Elon changed the goal post again after making that statement. His latest one is three years.

Source? Last I heard, Elon's wacky prediction was still "sleep in your car" by end of 2020.

 

[Saghost]

It is only Elon dreaming.
Elon changed the goal post again after making that statement. His latest one is three years.

With respect, no.

My point was that the whole Full Self Driving strategy has been based around a 50 TFLOPS computer running 8 cameras simultaneously using a deep neural net set trained by fleet learning - since sometime in 2016, Tesla's development team has believed that was what they'd need and how they'd get there.

In April they got the hardware for the first time - and they didn't change anything about the cameras when they did, so apparently they are quite satisfied with the second generation cameras of AP2.5. Now they have a fairly significant fleet of cars to do the learning against.

So this coming year is their chance to make the computers perform. Either they've got a viable plan and scope, and FSD will be ready next year, or the entire FSD development team and strategy has a critical flaw, and Tesla will have to re-evaluate everything, and possibly start over.

 

[DanCar]

Tesla's development team has believed that was what they'd need and how they'd get there.

With respect, no. Where you getting this from? The people I've heard from Tesla's development team, both directly and indirectly have a much different opinion. The ones that have told Elon the truth, get demoted or fired and the rest just stay quiet. Why do you think the previous heads of autopilot got let go? Jim Keller, Sterling Anderson, Chris Lattner. Stuart Bowers got demoted as previous head of AP software.

In April they got the hardware for the first time - and they didn't change anything about the cameras when they did, so apparently they are quite satisfied with the second generation cameras of AP2.5.

Quite a leap of faith, as will be proven wrong with future updates. Do you know the percentage of stop sign detection in the dark with current AP visualization feature? Close to zero based on one persons tests.

Now they have a fairly significant fleet of cars to do the learning against.

A lot of Tesla fans here drinking the Tesla kool-aid, and will only lose faith once it is proven. Doesn't matter that Tesla has been wrong several times on this subject already.

So this coming year is their chance to make the computers perform. Either they've got a viable plan and scope, and FSD will be ready next year, or the entire FSD development team and strategy has a critical flaw, and Tesla will have to re-evaluate everything, and possibly start over.

The chances of full self driving happening next year are zero, everyone with experience in the subject matter knows that except those that are drinking the kool-aid.
Drinking the Kool-Aid - Wikipedia

 

[DanCar]

Source? Last I heard, Elon's wacky prediction was still "sleep in your car" by end of 2020.

What is your source? He said in three years other cars will be like horses because of driverless capabilities.

 

[diplomat33]

What is your source? He said in three years other cars will be like horses because of driverless capabilities.

That's a different Elon quote.

My source? In Feb 2019, during the Ark Invest podcast, Elon said this: "My guess as to when we would think it is safe for somebody to essentially fall asleep and wake up at their destination? Probably towards the end of next year,"

By the way, I am not agreeing with Elon's quote. I am just giving you the source since you specifically asked for it. The fact is that Elon moves the goal posts constantly and makes crazy "forward looking statements" about FSD all the time. None of his FSD predictions should be taken seriously IMO. Heck, even the ark invest quote is riddled with "my guess" and "probably" implying uncertainty. So Elon is essentially making a wild guess when he says end of 2020 for "sleep in your car" self-driving.

 

[EVNow]

I think FSD will come to an inflection point in 2020.

Tesla has been building to this point for over three years, ever since they started really scoping the problem. They now have the computers they thought they'd need and a fleet of cars equipped with them to learn on.

So either they judged the problem correctly, and FSD will show up in 2020, or the Tesla experts misjudged the scope of the problem, and they won't have FSD for at least a few years, until they release a new set of computers and sensors (in which case there's a good chance someone else will solve the problems first.)

I'm not 100% sure which just yet.

There are multiple other possibilities. There is no compelling reason why these two are the only possibilities.

 

[DanCar]

That's a different Elon quote...

Here is the other quote, where Elon says in three years owning other cars will be like owning a horse.

 

[diplomat33]

Here is the other quote, where Elon says in three years owning other cars will be like owning a horse.

Yes, I remember that quote. So yeah, in one interview he guesses that we might be able to sleep in our self-driving tesla in 1 year from now and in another presentation, he implied that tesla cars will be so full self-driving in 3 years that our cars will be like horses in comparison. That's kinda my point: Elon is all over the map with his predictions.

 

[boonedocks]

I would be very happy if Tesla has Level 3 working on major expressways after entrance and before exits. I think it would give them a lot of breathing room to continue work and avoid further black eyes.