Apple abandons self driving car/says it’s not technologically feasible to achieve L5

[2101Guy]

Apple downsizes, delays its autonomous car plans: report — MarketWatch

Apple Inc. has scaled back its autonomous electric-car plans and delayed its target date until 2026, according to a new report.

stocks.apple.com

Apple’s Project Titan has been in a state of limbo at the tech giant in recent months, upon realization that the ambitious self-driving-car plan is not currently technologically feasible.

 

[diplomat33]

Apple downsizes, delays its autonomous car plans: report — MarketWatch

Apple Inc. has scaled back its autonomous electric-car plans and delayed its target date until 2026, according to a new report.

stocks.apple.com

Apple’s Project Titan has been in a state of limbo at the tech giant in recent months, upon realization that the ambitious self-driving-car plan is not currently technologically feasible.

To be clear, I don't think Apple is saying that L5 in general is not possible. They are saying they can't achieve L5. That is an important distinction to make.

But frankly, I don't know why people even pay attention to these stories about Apple's self-driving car. Maybe because Apple is a big and famous company? But the reality is that Apple's self-driving project has never seemed serious to me. Apple was never going to do L4 or L5. They only did 16k autonomous miles for all of 2021 and their disengagement rate was 1 per 20 miles. They have not released any details about their autonomous driving that I am aware of. All we have are fan concept art of an "apple car". So I am not surprised at all that Apple is rescaling to just do autonomous driving on highways. My guess is Apple will try to create a L2 "hands-free highway" or maybe "L3 highway" driving system to sell to automakers. They might even try to follow comma and develop some type of "L2 hands-free highway" kit that they could sell direct to customers. I could see that as a viable business model since Apple has millions of loyal customers who would likely buy a kit that makes their car L2 "hands-free".

 

[2101Guy]

Yes. I also interpret the L5 reference to be for their car.

Now that said, any thoughts on timeline for a L5 car from..anyone? L5 as in (at a minimum)NYC to LA with zero human input/intervention at all.

 

[diplomat33]

Now that said, any thoughts on timeline for a L5 car from..anyone? L5 as in (at a minimum)NYC to LA with zero human input/intervention at all.

Well not to argue the SAE levels but IMO, NYC to LA with zero input would not even be close to L5. That's because the car might do NYC to LA with zero input but require input on a trip from say NYC to Houston or Chicago to Orlando. Or maybe the car does NYC to LA on a sunny day with zero input but requires input to do NYC to LA on a rainy day. L5 means any trip, any conditions, must be zero input and safe.

Let's take a couple scenarios:

NYC to LA demo in good weather with zero input
I think Waymo, Mobileye or Cruise could probably do that demo this year. Again, as a demo. I am not suggesting it would be zero input every drive. I am just suggesting they could do one demo trip in clear weather that is zero input.

NYC to LA demo in bad weather with zero input
I think Waymo, Mobileye or Cruise could probably do a demo like that next year. Again, as a demo. I am not suggesting it would be zero input every drive. I am just suggesting they could do one demo trip in bad weather that is zero input.

Actual L5
I don't really believe in L5. I think we will get to L4 that works in so many areas and so many conditions that it will be close enough to L5 that we will essentially treat it as L5. For example, imagine a system that is driverless on 99% of US roads. It is technically still L4 since it is missing 1% of US roads but I think most people would probably consider it close enough to L5.

I think we are probably 5-10 years away from having the software that is "close enough" to L5. Waymo is probably the closest in terms of tech. But I think Mobileye probably has the best shot at actually scaling something that is "close enough" to L5. That is because of their scalable maps and their modular approach where they can add a radar-lidar stack to consumer cars that already have "L2 everywhere" to make them "L4 everywhere".

But ultimately, I don't think we will really care about L5. I think people will be happy with L4 that works where and when they need it to. So the goal will be to scale L4 that works for the most people in the conditions they need it. That's basically what Waymo (and Cruise) are trying to do. Waymo says they are trying to solve 4 domains: dense urban, suburban, freeway, and weather. Now these 4 domains might still not add up to L5 (it's missing rural and dirt roads) but I think it would still be very meaningful L4. I would even argue that those 4 domains would be "close enough" to L5. Imagine driverless that works in dense urban, suburban, freeway and weather. That would be L4 that could scale to a lot of people for a lot of useful driving needs.

 

[kabin]

Nothing wrong with waiting. I don't think the technology is there yet. Maybe at least one more design cycle and significant hardware advancement to reduce cost, size, and be more human-like on the road.

There's been dramatic advances since DARPA's urban challenge from 15 years ago but there's still a long way to go before there's human-like control, anticipation, and decisiveness. Until then AV vendors will continue to overstate capabilities of overpriced systems, develop patents, and spend money hand over fist.

 

[joelliot]

If you got back in Apple's history, they are almost never first. They where not first with 5G, Wifi 6, Graphical user interfaces, touch screens or pretty much anything else. What they are good at, is figuring out when a technology is ready for cost effecctive mass production and then jumping in. I would expect they would do the same thing with FSD. A couple of years after someone else get it barely working with expensive hardware, apple with jump in with an optimized solution.

 

[SourJellybeans]

I would bet it’s more that it’s not financially feasible than not technologically feasible.

(Which is not to say that I think L5 is coming from anyone, anytime soon)

Edit: I was also completely baffled by Apple getting into this space at all. Like, yah, you’ve got neat AR and ML tech, but being a car company is something totally different than categorizing a plant I took a photo of. That said - I was also baffled by Apple getting into the mobile phone space, and they showed me there.

 

[voyager]

The whole AV quest has turned into a self-serving 'industry', where thousands of engineers and AI experts have found
employment with in a lot of times spectacular salaries. It has been estimated that this 'industry' has burned through
$100 billion over the past decade. It was to be expected that investors would get impatient, particularly given rising interests.

 

[AndrewZ]

I would bet it’s more that it’s not financially feasible than not technologically feasible.

(Which is not to say that I think L5 is coming from anyone, anytime soon)

Edit: I was also completely baffled by Apple getting into this space at all. Like, yah, you’ve got neat AR and ML tech, but being a car company is something totally different than categorizing a plant I took a photo of. That said - I was also baffled by Apple getting into the mobile phone space, and they showed me there.

I don't know about the automobile hardware, but I think Apple is pretty successfully knocking it out of the park with CarPlay without even getting into their next iteration where they're working with carmakers to expand it to all screens and interface much more deeply than it does currently

Seriously have a hard time even considering a car today that doesn't have good CarPlay functionality. I want to plug in my phone and everything seamlessly flow, simple and straightforward. And then OTA updates for the car software come straight to your phone too, done and done.

The whole AV quest has turned into a self-serving 'industry', where thousands of engineers and AI experts have found
employment with in a lot of times spectacular salaries. It has been estimated that this 'industry' has burned through
$100 billion over the past decade. It was to be expected that investors would get impatient, particularly given rising interests.

Part of why I roll my eyes when people start bringing up AI experts opining and whose income relies on hyping things up, this is right around the corner, yadda yadda yadda. I wasn't born yesterday, follow the $$$.

 

[imwco]

Nothing wrong with waiting. I don't think the technology is there yet. Maybe at least one more design cycle and significant hardware advancement to reduce cost, size, and be more human-like on the road.

There's been dramatic advances since DARPA's urban challenge from 15 years ago but there's still a long way to go before there's human-like control, anticipation, and decisiveness. Until then AV vendors will continue to overstate capabilities of overpriced systems, develop patents, and spend money hand over fist.

This is true, think it's strange that the industry hasn't gone modular in it's AI design though. Seems like you could get 3-5 different companies serving the engineering use cases.

 

[diplomat33]

This is true, think it's strange that the industry hasn't gone modular in it's AI design though. Seems like you could get 3-5 different companies serving the engineering use cases.

I think they kind of are doing that. There are companies who are only doing autonomous small delivery, some are only doing autonomous semi-trucks, some are only doing autonomous campus shuttles etc... It's just that the companies like Waymo or Cruise who are doing full self-driving cars, get most of the attention.

 

[beatle]

But frankly, I don't know why people even pay attention to these stories about Apple's self-driving car. Maybe because Apple is a big and famous company?

I'm no Apple fanboy (I own no Apple products) but they do execute well and they have more cash on hand than any company on the planet to put into it. Anyone can say, "I want to make autonomous cars!" and raise a few bucks, but Apple doesn't need to do that. They just need to execute.

My guess is Apple will try to create a L2 "hands-free highway" or maybe "L3 highway" driving system to sell to automakers. They might even try to follow comma and develop some type of "L2 hands-free highway" kit that they could sell direct to customers. I could see that as a viable business model since Apple has millions of loyal customers who would likely buy a kit that makes their car L2 "hands-free".

An aftermarket addon like Comma, or a license to automakers sounds a lot more feasible. Both Mobileye AP1 and Comma seem to have pretty good systems for highway and traffic use. That's all I really care about anyway. I couldn't care less about how well a system will navigate city streets because I have no motivation to use it there.

 

[diplomat33]

I'm no Apple fanboy (I own no Apple products) but they do execute well and they have more cash on hand than any company on the planet to put into it. Anyone can say, "I want to make autonomous cars!" and raise a few bucks, but Apple doesn't need to do that. They just need to execute.

Yeah. I am not saying Apple can't do self-diving, I'm just tired of the rumors about an Apple self-driving car with no concrete product yet. And the CA DMV report has shown that Apple has one of the worst disengagement rates in the industry. So I don't put much stock in rumors until I see something real.

An aftermarket addon like Comma, or a license to automakers sounds a lot more feasible. Both Mobileye AP1 and Comma seem to have pretty good systems for highway and traffic use. That's all I really care about anyway. I couldn't care less about how well a system will navigate city streets because I have no motivation to use it there.

I think an aftermarket add-on, like Comma, would be Apple's best bet. It would be cheap for Apple to make. Apple already has some of the best computer chips in the world and certainly the engineering talent to do the software. And Apple has a big customer base that would likely jump at a chance to buy an Apple system that allows them to do "hands-free" driving in some cases. Apple could also sell the add-on directly to automakers who want an easy ADAS solution to put in their cars. I am sure Apple could find OEM buyers. And it does not need to be full self-driving so it would be much easier to do than trying to do L4.

 

[replicant]

Apple Ramped Up Autonomous Vehicle Testing Last Year, Filings Show

Government records show Apple’s secretive auto division almost quadrupled its self-driving miles in 2023.

www.wired.com

 

[diplomat33]

Apple Ramped Up Autonomous Vehicle Testing Last Year, Filings Show

Government records show Apple’s secretive auto division almost quadrupled its self-driving miles in 2023.

www.wired.com

Looking at the CA DMV report, Apple did 452,744 autonomous miles and had 3,194 disengagements. That's 142 miles per disengagement.

Here is the breakdown of the Apple disengagements:

DESCRIPTION OF FACTS CAUSING DISENGAGEMENT	Count of DESCRIPTION OF FACTS CAUSING DISENGAGEMENT
A discrepancy in the onboard map resulted in a motion plan requiring disengagement by the Safety Driver. The Safety Driver was able to safely take control of the vehicle.	279​
The ADS generated a motion plan that would have placed the vehicle in an incorrect location on the roadway, requiring the Safety Driver to disengage. The Safety Driver was able to safely take control of the vehicle.	91​
The ADS incorrectly detected the state of a traffic signal, which resulted in a motion plan requiring disengagement by the Safety Driver. The Safety Driver was able to safely take control of the vehicle.	4​
The ADS incorrectly predicted the behavior of another road user, which resulted in a motion plan requiring disengagement by the Safety Driver. The Safety Driver was able to safely take control of the vehicle.	1926​
The ADS initiated a disengagement due to a diagnosed hardware issue. The Safety Driver was able to safely take control of the vehicle.	667​
The ADS initiated a disengagement due to a diagnosed issue with the motion plan. The Safety Driver was able to safely take control of the vehicle.	136​
The ADS developed a motion plan to remain stationary, requiring disengagement by the Safety Driver. The Safety Driver was able to safely take control of the vehicle.	18​
The ADS incorrectly perceived an element in the surrounding environment, which resulted in a motion plan requiring disengagement by the Safety Driver. The Safety Driver was able to safely take control of the vehicle.	62​
The test vehicle sensor field of view was obstructed, preventing the adequate perception of the surrounding environment and requiring disengagement by the Safety Driver. The Safety Driver was able to safely take control of the vehicle.	11​
Grand Total	3194​

 

[diplomat33]

This might be a dumb prediction but I could totally see Apple doing a Comma.AI-like product. Call it iPilot and it is a small box device with 2-3 cameras and a computer inside that you can install inside your windshield and it allows your car to do lane keeping, cruise control and lane change on the highway hands-free but eyes-on. Maybe it is compatible with CarPlay too.

 

[kabin]

Rumors say Apple's EV launch with Level 2 is pushed out to 2028 from 2026. Maybe the plan is to let everyone else spent time and money figuring out what works best before diving in?

 

[joelliot]

Rumors say Apple's EV launch with Level 2 is pushed out to 2028 from 2026. Maybe the plan is to let everyone else spent time and money figuring out what works best before diving in?

I had mentioned up thread, but Apple is almost never first with any technology. They wait until someone else figures it out and then they copy that, sometimes so blatantly, they get dinged for infringement. …but that is another story.

 

[diplomat33]

Rumors say Apple's EV launch with Level 2 is pushed out to 2028 from 2026. Maybe the plan is to let everyone else spent time and money figuring out what works best before diving in?

Yeah, It will be 2028 at the earliest. So Apple definitely won't be first. Heck, there are already quite a few EVs out now and also quite a few L2 systems (Tesla AP, GM Super Cruise, Ford Blue Cruise, Xpeng NGP, Toyota Highway Teammate, Hyundai Highway Driving Assist, Lucid Dream Drive, Polestar Pilot Assist, Mobileye SuperVision, etc) So assuming Apple really does launch an EV with L2 in 2028, the real question will be what will it be like and how good will it be?

 

[enemji]

Really speaking, L5 cannot and should not mean it can replicate the best ever human driver. In other words, the vehicle should be able to stop driving when conditions are not OK for it’s sensors and computers, and restart when conditions are OK for it.