Tim Cook Confirms Apple's Focus on Autonomous Driving Systems

[DaveOP]

How about detours? Perhaps the have to build some temporary roads to move traffic around roadworks. That sure is no easy task for a self driving car. For human driver that is no problem, but for a computer program controlling the vehicle using

The problem is, you're thinking of self driving cars in the context of today. There are some interesting traffic studies out there about how much this will change patterns and road usage. When cars can talk to each other, you won't have as many random slow-downs, gapers delays, etc. Cars can also be smarter about parking, not needing to cruise around for spots and congesting downtown areas. They are not susceptible to drunk drivers, tired drivers, angry drivers, bad judgement. Sure, it will take years, maybe decades to get this stuff perfect, but it is the future.

 

[Denmac1]

I think so and it makes more sense than building and selling their own car.



That's what I'm talking about.
Car makers already have the factories and most importantly the network of car dealers and authorised garages where you can have your car fixed during warranty.
You can't take your car to the AS and have it fixed, and I don't think Apple is willing to open hundreds or thousands of garages

I don't believe Apple will manufacture their own cars. As they have done with Foxconn, they will source out the final product.
Once they develop the 'heart' of the systems, they will get someone like Magna Steyr to do the hardware. Then there will be service centers that will be already in place to take care of any warranty problems.

 

[JOLoughlin]

To be honest, I’d be dissapointed if they didn’t build an autonomous electric car and just built a “system” (whatever that means). I know it is a monumental task and there is strong competition but if they can’t combine hardware and software I don’t think they should bother in this area.

 

[AFEPPL]

Sorry I can't afford a 100k plus car lol wish I could. I have a lowly S3. But the tech even on BMW...yawn.

I would hope yours had something better to go along with the price tags.

The newer cars all get an option i believe for the VC. I'm not being an arse, it is a good system.

Audi MMI I'd pretty "cool looking" if you are into what tech looks like in movies - in reality, it is a pretty poor UX.

I disagree, it's "alot better" put together system vs CarPlay which i could also use too now on the Audis but "choose" not to. Audis VC is more logical, flows better and is better integrated in the most. Yes some option are deep in the system, but those that are not ones you play with while moving. Also Like BMW the control wheel with touch pad is just a more seamless way to interact with the system when in motion vs touch screen that just inaccurate and difficult.

 

[nt5672]

BS. I know people still using an iPad 2 and using Mac laptops that are 5 or more years old. Product obsolescence is due to the rapid pace of technology improvements not because Apple is shipping poor quality hardware that doesn't last.

Can you use old hardware? Sure, but not if you want the latest security patches. And not if you want the same performance as your device when it was new. And not if you want to use the latest functionality in Apple's own apps.

Apple has a strategy to make old hardware slow and out of date and provides no easy way to go back to previous system after you have found out that the latest system is dog slow and pathetic.

Quality has to do with things like having Safari in beta (technology preview) for over a year, with having to issue beta releases every 2 weeks, with having to release updates to production releases within weeks to correct released bugs, it has to do with the number of unfixed bugs that exist in every Apple product.

Now, lets talk about AI. Have you ever really tried to use Spotlight? It sucks. Have you every tried to use Siri? It sucks. The two prime example of Apple's use of AI are so bad that people laugh and just put up with them. Now translate that AI performance to an autonomous car. Well I won't at least!

Like I said, Apple does not have the business process in place to produce mission critical software what has to work all the time every time. Can they do it for email, twitter, reddit, etc., yes. But in a car?

 

[Glassed Silver]

At least when I say I want machines to make traffic safe I'm implying that I want the system proven to work. To me that means demonstrated performance and response to hazardous situations, not a theoretical improvement. I know a machine can be more attentive than I can be on long trips or times when I'm not at my peak. I know this is true of other drivers as well. I also know that people are very afraid of surrendering control over their safety to a machine (even though they do it all the time) so the first autonomous cars will necessarily come as high end options that allow people to become familiar and comfortable with the technology.

It's been happening slowly over a decade, and the first fully autonomous vehicles will merely be an evolutionary step in driver assist technology.

That's exactly what I want it to be, assistive. (read: optional, always able to balance my trust in my own ability to drive versus what the machine might give me in addition or when I cannot drive, but need "someone" to take over)

Personally I'd not use it for full auto pilot methinks, but to know that my car can avoid accidents that are not humanly preventable would be VERY welcome.

Ultimately it's not going to be anyone's "feelings" on the matter that drive adoption of this tech. At first, it will be the dramatic decrease in insurance premiums that owners of self-driving cars will pay (a poll found that acceptance of this tech increased dramatically when drivers were offered a 90% discount on insurance). Eventually it will be government policy requiring automation to use certain infrastructure (ie self-driving car only lanes). At some point it will be the default.

Oh it's not like I don't have an idea how it's going to happen, doesn't change the fact I don't like it.

Not trying to convince anyone to agree with me though.

Glassed Silver:win

 

[nt5672]

. . . .

There's a reason this is part of the macOS EULA:
E. YOU FURTHER ACKNOWLEDGE THAT THE APPLE SOFTWARE AND SERVICES ARE NOT
INTENDED OR SUITABLE FOR USE IN SITUATIONS OR ENVIRONMENTS WHERE THE FAILURE OR
TIME DELAYS OF, OR ERRORS OR INACCURACIES IN THE CONTENT, DATA OR INFORMATION
PROVIDED BY, THE APPLE SOFTWARE OR SERVICES COULD LEAD TO DEATH, PERSONAL INJURY,
OR SEVERE PHYSICAL OR ENVIRONMENTAL DAMAGE, INCLUDING WITHOUT LIMITATION THE
OPERATION OF NUCLEAR FACILITIES, AIRCRAFT NAVIGATION OR COMMUNICATION SYSTEMS, AIR
TRAFFIC CONTROL, LIFE SUPPORT OR WEAPONS SYSTEMS.

I know it's "standard stuff", but still. Software for self driving cars has to adhere to different rules. Not so easy to weasel out of liability.

Apple cannot even handle controversial political issues or freedom of speech because of how they feel it will effect their brand. What happens when a bunch of people die because of their software. Do you honestly think they will stick around and defend their brand. While we can't know exactly how they will react, we have some pretty good ideas based on past practices.

 

[BurgDog]

This is gonna be hard choice,
- On the one hand the Google car will follow the best directions to ge me there
- but on the other hand the Apple car won't get a virus and crash
[doublepost=1497356406][/doublepost]

Actually most planes land at airports with AI nowadays

I strongly doubt that AI of any sort is being used in those systems. Landing systems are fairly simple compared to cars on public roads, the software used is very well tested, the systems highly redundant, and the pilot is always monitoring and able to intervene if needed.

 

[CTHarrryH]

I really would love to know who really wants these cars? I know of no on who wants one - they might like to use the feature occasionally (when in heavy traffic or a long trip on the interstate) but not full time and not one that doesn't let the driver take over which is what is being aimed for. How will it know that one side of the street requires coins for meter parking but the other takes credit cards - where will it park.
The other total unknown and will be a critical issues is cost. They will cost more than current cars but how much it won't be cheap. Plus how much will the monthly subscription for network access and map updates cost? If the 3 year cost of ownership is 150% of a normal car who will buy?
I can see for freight carriers or similar but for mass quantity every day drivers I don't see it.
[doublepost=1497365735][/doublepost]

Use a little imagination. So let's say the car is limited to 65 MPH on highways and 30 MPH in the city. On the highway there will be much less congestion because the cars are all talking to one another and coordinating their acceleration and deceleration. In the city there will be less waiting at stoplights as the infrastructure itself will regulate the flow of traffic based on demand. This will save a huge amount of time. Fewer accidents means less congestion. Plus you get to actually spend the time in the car being productive (or lazy) instead of stressing out driving.

Now compare it to going an unsafe 85 MPH on the freeway on your own. There is a much higher likelihood of an accident, which could be fatal at that speed. And then you cause a huge traffic jam behind you as they remove the charred vehicle from the roadway. For the potential savings in time of a few minutes as most vs going 65.

And eventually, as self-driving tech improves, you'll have "trains" of cars going 100 MPH inches from each other, all in constant communication and coordinating every move they make. This will be dramatically faster than the current setup.

And who is going to pay for the infrastructure needed to coordinate lights, etc. Who is going to pay for the 50% of the cars who can't communicate with each other?

 

[stanman64]

I'd bet my ass that Cook drives a Tesla

 

[jayducharme]

Would you be riding at 80 MPH on a software?

Only if all the other cars on the road were also powered by software. I'm not so much afraid of the computer failing; I'm much more afraid of bad judgement on the part of a human driver. Remember that in over a million miles, Google's autonomous car has had very few accidents. And those that it had were predominantly due to human error.

As far as Tim's comment about electric vehicles being fun to drive, I can vouch for that. I leased one for three years (a Smart EV) and it was the best car I've ever had, perfect for getting me back and forth to work each day. The cost of battery replacement -- that was the big issue and was the main reason why I didn't buy the car when my lease expired.

 

[applezulu]

A few thoughts:

To those who are touting their own stellar driving experience and many miles driven without an accident as a reason self-driving cars shouldn't be developed... It's great you're a good driver. Me too. We're above average! Guess what? Half of the drivers out there are above average, and except for in Lake Wobegon, the other half are below average. All of us out there behind the wheel start on a learning curve, accumulate experience, most get better and better, until they peak, and then as we get older, response times slow, and we get worse and worse, until we plow into some pedestrians or someone in the family sees it coming first and takes the keys away.

Software isn't infallible, and neither are you. I'm not, either. At some point, well-developed software will get better than the average driver (it might already be there) and will keep getting better. As more and more of these cars are first tested and eventually put out on the road for consumers, they will cumulatively accumulate experience at a rate of millions of miles driven per year. No matter how good and how experienced a driver you are, you won't be able to match the level of experience accumulated by the AI built into some future Apple car. Unlike a blank-slate fifteen-year-old human with a new learner's permit, every new self driving car will come off the dealer lot with millions of miles of driving experience.

It's also pretty inevitable that this technology will employ better means of driver-to-driver communication than a blinking light, a horn, eye contact, or 'the finger.' So the AI will accumulate more experience than is possible for you or I, plus it will be able to communicate directly with other AI drivers using far more data points than you or I can convey.

In one other related thought, it's important to realize how self-driving cars will be a game changer for all of us as we get older. Most people don't live in urban areas sufficiently served by public transportation. We they get old enough that they shouldn't be driving any more, the loss of independence can cause their lives to start on a dramatic downward spiral. Others have to take them places or bring things to them when they can, which is probably not when the older person wants or needs it. Social lives are diminished and the inability to get out and do things hastens physical and mental decline. Other options include moving to various levels of elder care living places, where independence is still pretty limited and you're constantly surrounded by old people waiting to die.

It will be a revolutionary change when an average 90-year old can get up in the morning, slide into the Apple car and go do whatever they want with others of all ages, like anyone else. They'll be able for much longer to continue living a life at home without being a burden on others. That is going to be incredible.

 

[BurgDog]

You do realize that tens of thousands of planes traveling near the speed of sound are piloted 99.9% by software every day, don't you?

No they're not. Every aircraft has a pilot in command who is in control at all times. The software assists are just helpers no more than speed control on cars.

 

[neutralguy]

I'm curious as to why advanced control and measurement techniques nowadays are called AI. There's really not much intelligence included in keeping a car in a lane or avoiding obstacles. It's just programmed reactions to input from sensors.

I didn't say it was easy, just that it is not really that much "AI" included. I don't think anyone would call the system that controls your heat pump at home "AI", it just keeps the room temperature at your desired setting with regards to the outside temp, inside temp and perhaps some other variables. Reactions to sensor input according to algorithms.

The interesting part will be when the elephant in the room has to be discussed: ethics.
A self driving car might have to make very tough decisions like:
To avoid the crash going straight ahead, killing the occupants in the car, the other options are to kill the pedestrians to the right or the cyclists to the left but saving the occupants of the car.

A human driving a car on the right side of the road would probably weer to the right killing the pedestrians as a reflex, but the car with "AI" has to make a "decision".
None of the options are the obvious right one. The self driving car cannot "think", it will do what the algorithms tells it to do.

A car with "assistive systems" may help the driver in whatever decision he or she makes, but the car does not make the decision.

When the car makes the decision, is the driver no longer at fault?
-"But officer, the car ran over that old lady by itself, I was looking at Facebook on the screen".

Or will the driver be responsible for whatever actions the car takes, even if the driver has no control?

I see some very interesting court cases up ahead...

^this

The more I look at self driving cars, the more I feel like it's a fad similar to 3D tv. Nobody wants it. Company wants to push it.

The only way self driving cars will work is when one fine day you wake up and see all the cars are replaced by self driving cars. Until then, it's gonna be very frustrating for people to be on self driving cars.

Also, why do we want self driving cars? Is it because of traffic congestions or because we want to use that time to browse Facebook? May be companies should invest all the money and minds to build better public transportation system and all our traffic problems will be solved.

 

[DotCom2]

This still doesn't tell me if they are making a physical car or just the software to run a car when they say "We're focusing on autonomous systems".

 

[vladi]

Ivy said the same thing about the Watch and look how that product turned out. Money doesn't guarantee you anything in this world and age.

 

[newdeal]

It makes sense if Apple succeeds and you look back on it the way we now look back on the iPhone.

But if you're living in 2005, an "iPod Phone" is just Apple trying to get into an industry that it knows nothing about and in which has no chance in succeeding

.

Its not the same, because the phone can be sold with 40% margins. The car isn't going to have good margins at all, it will also require a network of locations where you can repair it, stocking parts for years for past models etc etc. Apple could surely do it, but making a car isn't high enough margin for them to waste their time. Putting systems in other manufacturers cars makes sense on the surface but I don't think the manufacturers will really want to do it as I think they will want full control over any AI system

 

[konqerror]

Machines will fail, certainly. But humans fail at a much higher rate.

Actually, when you look at the US statistics, humans get in a fatal accident once per 100 million vehicle-miles, or around 4 million vehicle-hours. This is a very high bar for a machine. For example, a MTBF of a computer is maybe 50,000 hours. So if the failure of a computer leads to a fatality, AI driving is very unsafe. (Aviation computers run into this. They require triple redundancy, dissimilar hardware designs, very expensive software development techniques and they don't use AI at all)

Imagine if the Apple car software beachballed when approaching an intersection.

The reason why you perceive driving to be unsafe is simply the sheer number of vehicles and miles traveled results in large number of deaths. (Again in aviation, extrapolating increasing demand with current safety rates, in a few decades we'll be crashing one entire jet a week)

 

[FrenchRoasted]

The problem is, you're thinking of self driving cars in the context of today. There are some interesting traffic studies out there about how much this will change patterns and road usage. When cars can talk to each other, you won't have as many random slow-downs, gapers delays, etc. Cars can also be smarter about parking, not needing to cruise around for spots and congesting downtown areas. They are not susceptible to drunk drivers, tired drivers, angry drivers, bad judgement. Sure, it will take years, maybe decades to get this stuff perfect, but it is the future.

^this

The more I look at self driving cars, the more I feel like it's a fad similar to 3D tv. Nobody wants it. Company wants to push it.

The only way self driving cars will work is when one fine day you wake up and see all the cars are replaced by self driving cars. Until then, it's gonna be very frustrating for people to be on self driving cars.

Also, why do we want self driving cars? Is it because of traffic congestions or because we want to use that time to browse Facebook? May be companies should invest all the money and minds to build better public transportation system and all our traffic problems will be solved.

The rest of us who have gotten over the thrill of everyday driving (we're well north of our 20s) want this. Most of the US doesn't have the density to have the same rail systems of EU or Japan so automated driving is a good solution. I want to be able to go out and drink adult beverages at a restaurant and not have to worry about driving home. AI can handle other bad drivers than I can, mostly because it won't be susceptible to road rage.

 

[applezulu]

... Nobody wants it. ...

Also, why do we want self driving cars? Is it because of traffic congestions or because we want to use that time to browse Facebook? May be companies should invest all the money and minds to build better public transportation system and all our traffic problems will be solved.

Why? Because many people out there can't, don't want to, or shouldn't drive. See my comment above about the significant positive thing self-driving cars will be for people as they get older. There are also any number of other circumstances where people can't or shouldn't drive.

As for "don't want to drive," I am a good driver and enjoy driving. I don't like commuting, however. I think I'd be happy to cede that task to a reliable autonomous vehicle. The same goes for the middle part of long road trips.

There was lots of harrumphing, right after Steve Jobs introduced the iPhone, about how nobody would want a smartphone with no physical keyboard, too. Just because you don't want one doesn't mean nobody wants one. Plus, there's a pretty good chance you will want one eventually. We all get old if we don't die first.

 

[drewyboy]

Its funny to read this, if you replace "car" with "phone".

It's funny to read this, if your replace "car" with "social network". Oh wait...

Way to cherry pick one of their handful of major company successes. How about you replace it with any of their failures or "hobbies" **cough cough apple tv**
[doublepost=1497368852][/doublepost]

Actually, when you look at the US statistics, humans get in a fatal accident once per 100 million vehicle-miles, or around 4 million vehicle-hours. This is a very high bar for a machine. For example, a MTBF of a computer is maybe 50,000 hours. So if the failure of a computer leads to a fatality, AI driving is very unsafe. (Aviation computers run into this. They require triple redundancy, dissimilar hardware designs, very expensive software development techniques and they don't use AI at all)

Imagine if the Apple car software beachballed when approaching an intersection.

The reason why you perceive driving to be unsafe is simply the sheer number of vehicles and miles traveled results in large number of deaths. (Again in aviation, extrapolating increasing demand with current safety rates, in a few decades we'll be crashing one entire jet a week)

You do realize that cars currently ARE run by computers right? When's the last time one of those failed on you?

 

[masterbaron]

fourthtunz said: "So this is why we can't get a good Mac? Cook is too busy using iPads and dreaming about self driving cars.
No one is minding the store ;("

There ya go again ... so true, all that wealth and all that emphasis on quality (back in the day) and they won't give us "deep skills" as it relates to home integration, migration path, upgrade ability, battery-life instead thinness, licensed software (clones), backwards compatibility, ports, etc. all in the name of profit.

But really, I mean really at what point do people realize too much of this stuff is simply making us dependent and vulnerable to attack both externally and internally as a woman using her iPhone walked right into a cellar door this week and fell 6 feet seriously injuring herself.

and when exactly was the last time someone hacked into your head?!

 

[applezulu]

Of course, it's important to realize that autonomous vehicles are not even a new concept. You can be pretty certain that this horse knew the route.
[doublepost=1497370108][/doublepost]

...

and when exactly was the last time someone hacked into your head?!

When was the last time you bought anything in a grocery store? The smells, sounds, and placement of everything in your average grocery store is a total psychological hack. When is the last time you clicked on an internet link? Just one weird trick? When was the last time you found yourself doing something for your significant other that you hadn't originally planned to do? I won't even get into a number of different inadvisable votes inexplicably cast by large numbers of people in recent times. Human heads are the most hackable devices out there.

 

[oliversl]

Tesla will get it 1st, Waymo will fail as its a moonshot project, Uber will come later, all manufacturers later and then Apple will enter the space. Meanwhile, our iOS will continue to crash, Siri does not understand us and the Macs are getting more expensive with less ports and power every year.

 

[bollman]

Actually, when you look at the US statistics, humans get in a fatal accident once per 100 million vehicle-miles, or around 4 million vehicle-hours. This is a very high bar for a machine. For example, a MTBF of a computer is maybe 50,000 hours. So if the failure of a computer leads to a fatality, AI driving is very unsafe. (Aviation computers run into this. They require triple redundancy, dissimilar hardware designs, very expensive software development techniques and they don't use AI at all)

Imagine if the Apple car software beachballed when approaching an intersection.

The reason why you perceive driving to be unsafe is simply the sheer number of vehicles and miles traveled results in large number of deaths. (Again in aviation, extrapolating increasing demand with current safety rates, in a few decades we'll be crashing one entire jet a week)

This is indeed a very valid point that I actually have overseen (baffled here).
Who hasn't seen any software hesitate for a second, unexplainable. At 70mph that can absolutely be fatal.

For those here who thinks that we will ever get to the point where cars are so autonomus that you can get home pissed "driving" your own car without any consequences: I hate to rain on your parade, but it will not happen, not in our lifetime at least. Wherever this trend may lead us, you can be certain that the operator of the vehicle (maybe not driver) will be responsible for what how the vehicle behaves in traffic.