Apple layoffs hit nearly 200 employees in self-driving car division

Originally published at:

The company says 190 employees in Santa Clara and Sunnyvale, CA will lose their jobs.


So, it’s going well then?

It’s just that it seems success begets layoffs as though it were SOP these days.


Depends if the people whose services were no longer needed were the developers or the drivers.


Well now what? Those cars aren’t just gonna dr…


Back to the old drawing board, I guess.


Yeah, I wonder what this means. It sounds like it was mostly engineers who were let go. They’ve re-focused the project once already, and this could be another such move. But I kind of feel like everyone’s realizing that self-driving cars are actually a much harder (and more distantly solvable) problem than has been acknowledged, and maybe they’re backing off of plans (at least more immediate ones, anyways). Or maybe they’re conceding the arena to Google, because reportedly Apple’s cars needed human intervention 10,000 times more frequently than Google’s. (Literally - once every 1.1 mile versus every 11,000 miles.)


I still think the smart move would be to introduce them in retirement communities before trying to get them to work on public streets. You’ve got all the right factors:

  • High need (many elderly can’t safely drive themselves)
  • Slow speed limits (reducing collision risk)
  • Private property (meaning fewer traffic regulations preventing deployment)
  • Limited range (fewer unexpected road conditions)

Plus if you run someone over you’ve cut their life short by, like, five years tops. :wink:


For some reason, my brain initially missed the word “layoffs,” briefly making that a much more exciting headline.


Come over to Detroit.

1 Like

Maybe they used the technology for the cars to create a self operating division?


Makes a lot of sense. They would just have to drive faster than golf carts to be useful there too.


Heck, even a self-driving golf cart could be a major convenience for people with mobility issues.


They don’t need them. Full self driving capability will be released by December 2018.


It’s always struck me as a long shot that Apple would make a car. They make very few core products, and they don’t release something new unless they are convinced that their version of that thing is a no-brainer choice (subject to price and taste). Or at least, that is how the Steve Jobs version of Apple operates, and that is still very much what they aim to be.

To my mind, the iPhone of cars would need to be something that, like, doesn’t even have a steering wheel. It can’t be some incremental evolution of what we have, because their business model is based on defining their own categories; that’s what lets them charge more and (more importantly, to them) dictate design choices to customers instead of the other way round.

But even if Apple has the wherewithal to make a radical car like that, it’s not clear there is a place in the world for it. Like, if you go on a family car trip, does everyone sit facing each other in a circle or something? We’re talking about a wrenching change to how society works, and that’s not going to be resolved on the first try. It’ll take a lot of blind experimentation and throwing silly ideas at the wall, which is the opposite of the kind of business Apple typically involves itself in.


Wouldn’t you need self-driving clubs and self-putting putters?

1 Like

“You’re all fired… You can find your own way home.”

1 Like

We already have many self driving cars; train systems, large farm equipment, large mining equipment. Large boats and planes are almost self driving. I think that the next step will be transport trucks on certain highways.
Also I think several steps will happen before self driving cars are built; cars will require wifi or some other radio transmission so that the police know where they are and speed they are going. To a certain extend this is already happening with private corporate fleets. And, it’ll probably happen in authoritarian places like Singapore first. All cars will get many more cameras. Rearview cameras are not becoming standard, soon regulation will make cameras on side mirrors mandatory, and a dash cam for recording accidents mandatory. Next, the road will also get smarter, the next generation of highway cameras for monitoring traffic will be better. Perhaps companies like Google should donate money for better traffic cams.

Once these three factors come into play; radio monitored cars, cars with many cameras, better traffic monitoring cameras. Will we see the next iteration of self-driving cars. This current iteration was possible by combining many existing technologies; fast computers, cheap radio technology, cheap radar technology.

1 Like

There was a period where that was true; self-driving cars have been possible for a few decades, if only there were machine-readable lane markers and traffic signs everywhere and so on. And it didn’t happen because no one was going to spend billions on that infrastructure before a single self-driving car was even sold.

The only reason it’s a realistic prospect now is that machine vision has got to a point where that infrastructure isn’t necessary. I mean, self-driving cars will have network connections because that’s table stakes for any gadget these days, but they won’t need remote monitoring to operate safely; or at least, they won’t go on sale until that is true. And no government will permit a self-driving car that can be made to disobey traffic rules. If remote monitoring and safety cameras are made mandatory, it’ll be to guard against human drivers.

Fully autonomous cars still imply major infrastructure change, but more as a reaction than a prerequisite. For example, once robocars are common, cities will start aggressively prohibiting on-street parking, as that’s a colossal waste of road, and if your car can drive itself, then it can park itself a mile or two away. In the very long term, a highway lane might look like an unmarked eight-inch strip of paving running through a meadow, with no signs or barriers, traversed by cars with in-line wheels forming aerodynamically efficient trains. As long as the robocars come first, authorities will have a strong incentive to transition to much more compact, efficient road systems that can’t be used by manual cars.

I see it slightly differently. The cargo industry is the one that so far has invested heavily into practical technology. Many large trucks all have GPS and are centrally tracked. The insurance industry would like to have dash cams installed on all cars. There is still lots of room to advance in these two industries, way more dash cams for insurance, way more tracking and network connectivity for trucking.
Meanwhile the hardest form of automation for cars is stalling, driving people around town, to work and bars.
Perhaps the transition step would be to have long haul trucks driven remotely. This would improve network capabilities, and govs would be encouraged to upgrade highway monitoring equipment by the industry, to make remote piloting a truck possible.

We don’t have “self driving cars” so much as “self driving cars*.” That is, cars that can operate autonomously in limited, non-street uses. It can’t recognize a baby in the street (not part of its training data), can easily be confused by street signs (any sort of alteration can make it be ignored/misread as something else entirely), throws a fit if it sees a plastic bag in the street, and is generally totally fucked by any amount of non-Californian weather (e.g. snow).

I mean, it’s already happened. New cars have internet connections. The smarter the car, the more surveillance is already being applied to it. It’s about corporate data collection rather than government surveillance, but the first is a big boon to, and encourages, the second. GM is collecting information about what people are listening to on the radio. Tesla is collecting information on… everything.

Autonomous car-to-car wireless communication is being planned, but comes with its own set of problems (e.g. making sure that bad actors aren’t injecting false information into communications to cause traffic accidents or advantage their own movement in some way, etc.)

Autonomous cars are also covered with cameras (and LIDAR, RADAR, etc.). That’s not the problem. The stumbling blocks are software-related. And there’s some of the same stumbling blocks as with autopilot systems in airplanes, despite that being a much simpler problem to solve (you don’t have to worry about collisions, street signs, road markings, etc. in the air). A major problem with airplanes is the overlap of machine and human control - a major cause of crashes is when the autopilot switches over to the human pilots when there’s a problem. As long as autonomous cars require humans to take control in certain situations, it’s not an autonomous car, whether that’s every 1 mile (as per Apple’s car) or 11,000 miles (as per Google’s). You can’t safely switch over to human control unless they’ve been waiting and ready the whole time. That might work for testing purposes (or not, as Uber’s completely-avoidable fatal accident showed), but it doesn’t work for driving purposes - that’s only usable for human piloted cars as extra safety features.