Apple Has Dismissed More Than 200 Employees From Project Titan, its Autonomous Vehicle Group (cnbc.com)

Apple has dismissed just over 200 employees this week from Project Titan, its stealthy autonomous vehicle group, CNBC reports. From the report: An Apple spokesperson acknowledged the layoffs and said the company still sees opportunity in the space: "We have an incredibly talented team working on autonomous systems and associated technologies at Apple. As the team focuses their work on several key areas for 2019, some groups are being moved to projects in other parts of the company, where they will support machine learning and other initiatives, across all of Apple," the spokesperson said. "We continue to believe there is a huge opportunity with autonomous systems, that Apple has unique capabilities to contribute, and that this is the most ambitious machine learning project ever."

Comments

Give It Up

[Anonymous Coward]

Drivers are required.

Re:Give It Up

[jwhyche]

This is what it really is. We are not ready for self driving cars, the infrastructure isn't in place to support them. For self driving cars to be reliable the roads will need to be specifically designed to support them.

Re:Give It Up

[Powercntrl]

For self driving cars to be reliable the roads will need to be specifically designed to support them.

And exactly what technological innovations are we lacking that prohibits this from being accomplished?

Your cell phone works because we've put up towers all over, you have electricity because we've run wires, GPS works because there are satellites in space, etc. We're not going to give up on self-driving technology just because some infrastructure needs to be put in place to make it work.

Re:Give It Up

[jwhyche]

And exactly what technological innovations are we lacking that prohibits this from being accomplished?

There is none. In fact with today's level of technology if we where to design and build roads for self driving cars it would be stunningly simple.

What I'm saying is the the current infrastructure in place we are not ready for self driving cars. We are trying to do this the hard way, which is to teach computer cars to drive like humans. With the current level of technology will not work. There are two many variables for the cars to analyze and do it safely 100% of the time. I can get in my car and drive two miles down the road and the road conditions will change completely. We are not ready for this.

The most ambitious machine learning project

[Anonymous Coward]

...this is the most ambitious machine learning project ever.

Wait until RealDoll perfects the small-talk subroutines for their A.I.-equipped "companionship" dolls. Simulating female bitchiness such as being able to make crazy links between totally unrelated events is going to require massive amounts of processing power.

Re:The most ambitious machine learning project

[TimMD909]

...this is the most ambitious machine learning project ever.

Wait until RealDoll perfects the small-talk subroutines for their A.I.-equipped "companionship" dolls. Simulating female bitchiness such as being able to make crazy links between totally unrelated events is going to require massive amounts of processing power.

Solving the problem of creating crazy links between disparate ideas has been solved for over a decade.

Re:The most ambitious machine learning project

[supremebob]

Hmm... Sounds like we should be expecting a #androidmetoo campaign in about 20 years when the fembots become self aware and sue their owners for abuse.

I think Apple is waiting to se the space shake out

[SuperKendall]

Apple I smart I think to not build a whole car, integration into car systems is the better idea for Apple.

I'm just not sure how Apple could bring anything extra to the driverless car space, which has quite a lot of players of all kinds feverishly working to deliver results.

Maybe after that area has solidified they might find some way that makes sense to enter, but for now sitting back and waiting is a good idea.

Ruleset

[fluffernutter]

Driving (anywhere on the surface of the Earth) cannot be defined with a reasonably finite set of rules, period.

Re: Ruleset

[artemis67]

Clearly it wasn't cancelled. More likely, Apple changed direction.

I suspect they initially wanted to build the whole car, but after several years of exploring the idea, it seemed impractical and not likely to yield the level of profit margins Apple is used to. So I think they reconfigured the project to offering just the self-driving hardware and software to existing car manufacturers.

I think at this point they've taken the research as far as it can go and, as you suggest, recognized the (current) insurmountable problems with full automation.

I think that a fully automated SDC would require either realtime AI processing or a significant infrastructure upgrade, or possibly even both. And the cost model doesn't seem to be there yet, as the sensors are still ridiculously expensive for the average consumer.

Re:Ruleset

[s_p_oneil]

"Don't hit things. Don't fall off a cliff." - unless they're birds that aren't afraid of your car, or rain/snow/fog is fooling your sensors

https://www.bloomberg.com/news...

“These things will not be functioning in anything heavier than a light rain”

I truly wish they would come sooner rather than later, and I support all of the testing and development they're doing, but it will still be quite a while.

Re:Ruleset

[MooseTick]

"Driving (anywhere on the surface of the Earth) cannot be defined with a reasonably finite set of rules, period."

So, why are we giving 16 year olds keys to cars with often only a few hours training?

And are you telling me an algorithm can't be created that simulates the sense nearly every animal in existence possesses? Sure, it may not work in rain, snow, or some conditions; but if it worked half the time it would be a boon to humanity.

Re:Ruleset

[fluffernutter]

Apparently after they wash the Waymo cars they have to get on top and wipe off every ... single ... drop so that it will work reliably.

Re:Ruleset

[fluffernutter]

Because we accept 16 as the age of driving because in our experience most 16 year olds understand all they need and are capable of driving. What I'm telling you is that there is no algorithm that can run on the car (because any wireless connection would be too slow) and identify that animal within the 0.001 seconds it needs to in order to make the correct decision.

Re:Ruleset

[drinkypoo]

What I'm telling you is that there is no algorithm that can run on the car (because any wireless connection would be too slow) and identify that animal within the 0.001 seconds it needs to in order to make the correct decision.

You do realize that in 0.001 seconds, a human literally cannot possibly even see that animal? If we saw anything, it would just be a blur. And we certainly wouldn't have time to think about doing anything about it.

Re:Ruleset

[s_p_oneil]

LOL. Maybe they could create a tiny little robot arm next front of each "eye", and when you wake the car up, it could "rub its eyes".

Re:Ruleset

[goose-incarnated]

"Driving (anywhere on the surface of the Earth) cannot be defined with a reasonably finite set of rules, period."

So, why are we giving 16 year olds keys to cars with often only a few hours training?

They drive better than the current crop of AIs.

Re:Ruleset

[fluffernutter]

If, by what you mean as reaction time, is the time to make a decision to react using on board sensors and processing, I am not disputing that a computer will be faster no contest. However, if you actually read what I wrote, I am concerned about the time it must take the car to make a decision. Is Google even close to recognizing a *still picture* of a duck from a dog from a cat from a beaver using their entire datacenter within the time the car will need? how does this datacenter get fit on board to the car so that this decision can be made in real time? Keep in mind that there are thousands of samples a second; so it will need to search through it's entire database of 'world objects' at least one thousand times a second, on board, in a car. Good luck. Just try not to hit or kill anything please.

Re:Ruleset

[dgatwood]

Is Google even close to recognizing a *still picture* of a duck from a dog from a cat from a beaver using their entire datacenter within the time the car will need?

If it is flying in the general direction of your windshield, I don't care if it is a duck or a dog or a cat. The car had better stop. Being able to identify objects is not necessary. What is necessary is the ability to classify objects as something that will cause damage when hit or not, along with the ability to determine whether those objects have the right of way or not.

Re:Ruleset

[fluffernutter]

Which they can't do either.

Re:Ruleset

[dgatwood]

True, but it is orders of magnitude less complex if the only things you care about are how the object is moving (i.e. does it have significant mass), whether it is big, and whether it is a person.

Not Surprising

[TexasDiaz]

I bet the employees that were laid off were part of a company they bought to specifically address autonomous driving, and no employees that were actually hired by Apple itself to work on Apple-related projects. Such is life when your company gets bought out by a monolithic entity, at that point you're pretty much expendable. Sure, I'm betting some (like software devs or hardware experts) were able to shift, but not everyone's roles are that movable.

Re: Not Surprising

[Type44Q]

There's at least a fifty percent chance that you're 100% wrong.

Re: It's OK, though

[PPH]

And then sue all the other car makers for using rounded corners.

Re: I think Apple is waiting to se the space shake

[cormandy]

Donâ(TM)t forget planned obsolescence so you have to replace your car every few years. Or how about really thin cars so you have no choice but to buy a case for it if you are to survive an accident.

No iCar

[AndyKron]

Good. I don't want an iCar

Re:No iCar

[aberglas]

Indeed. The iCar that only runs on iRoads might be somewhat limiting.

Re:I think Apple is waiting to se the space shake

[bill_mcgonigle]

In the distant past Apple would pay people to work on new ideas in case they came up with something new they could sell - the ATG. Most of Apple's new tech came from there and they made money commercializing those ideas. They trimmed non-product research when they had money problems. Steve claimed that all necessary research would happen within the product groups, but he knew well that kind of R&D doesn't generate new markets - this was reasonable as he was cutting the company back to four products, but it wasn't until the iPod group came onboard that they had something new.

Does Apple do anything new these days? Every product release seems to be a slightly faster, slightly thinner, slightly bigger-or-smaller version of the same products they've been selling since Steve was a vegetarian, from what I've seen.

Whether or not they're the richest company in the world this week, it wouldn't kill them to try some new stuff. Unfortunately they no longer have a leader, they have an efficient manager in charge. Does Tim even trip?

Need to look harder...

[SuperKendall]

Does Apple do anything new these days?

The letting go of people from this car group is proof they do.

Just because this did not work out does not mean they do not have other ideas under way that might.

FaceID, the Apple implementation, was very new (who else was doing 3D face auth in mobile profile before that) and just was released last year... cut them some slack man! Can't have something new and amazing every year.