A Day With Project Ara, Google’s Crazy Modular Phone




Seth Newburg has a phone. It’s a new phone, just a prototype. Almost no one here at Google has seen it yet. Not even Sundar (as in Sundar Pichai, Google’s new head of nearly everything). Which is just bonkers, because Sundar sees it all—especially important phones like this one. But Sundar’s meeting got delayed, and so now Newburg is sitting in a corner conference room in Google’s Mountain View headquarters, fresh off an airplane, with me at his side, and a photographer standing over his shoulder firing through pictures of this brand new phone: a prototype that’s unlike anything anyone has ever seen before. This is a future-of-Google phone. Hell, it’s a future of all phones phone. If things go well.


But so far they haven’t gone that well, at least not publicly. Newburg is trying to boot up a Project Ara prototype. Project Ara is Google’s dramatically boxy modular cell phone, and this is the final version of its Spiral 1 prototype. It’s a revolutionary concept: Phones should be made entirely of components that their owners can swap out. Not only will that give them a much longer life—you’ll be able to upgrade just your camera, say, or only the processor—but it also will accelerate development. You’ll get cheaper phones, better phones, more environmentally friendly phones. It’s a trifecta. So Ara has generated intense interest.


But it’s also fallen on its face a couple of times. The last two times Newburg, the project’s principle engineer, tried to boot up the phone in public were, well, bummers. In the first outing it had a broken display (which led people to make the obvious joke that one makes about a modular phone designed around swappable components but that can’t boot because the display is broken). The second time, at Google IO, it sort of kind of booted, and then it froze up. It got to the screen where you see the word Android, at least. But it didn’t actually come alive. Thankfully, that demo was followed by the latest Spotlight Story, a heartwarming animation by Glen Keane called Duet that made pretty much everyone cry and forget about the phone not booting. (Well, not everyone, but at least people left talking about Duet.)


Today, I’m at Google to see it try again. If it works, it will be the first time an outsider has seen the phone actually boot. Which puts a lot of pressure on these guys. It sucks to fail in front of the press. Again.


In fairness, this is really hard stuff. You can think of Ara a little bit like a Lego phone. Distinct modules, each with its own purpose, jack into an endoskeleton frame and snap into place with electro-permanent magnets. One for the battery. Another for the display. Yet another for the camera, and the antenna, and the LED light. Maybe even one for your blood glucose monitor. A blood glucose monitor? Well, sure. Why not. Project Ara is like old-fashioned 1980s-era Southern California punk rock music; it is whatever you make it to be.


Paul Eremenko, who formerly ran a modular satellite program at Darpa, now runs Google's modular phone project, Project Ara.

Paul Eremenko, who formerly ran a modular satellite program at Darpa, now runs Google’s modular phone project, Project Ara. Talia Herman/WIRED



“This isn’t rocket science,” says Paul Eremenko, who, before coming to Google, was a rocket scientist. Now, he runs Project Ara for Google’s Advanced Technology and Projects division. ATAP is like a Google within Google. It’s modeled on Darpa, and specializes in taking extremely challenging, almost impossible projects, and cranking them out within two years. Fittingly, Eremenko used to work at Darpa, where he ran the Tactical Technology Office. Among other things, he oversaw something called System F6, a “fractionalized spacecraft” program. System F6 took the functions that are today performed by one large satellite, and split them up between a series of smaller, modular satellites that traveled together in tight formation. Sounds familiar, right? It was very much like an outer space version of Ara.


Maybe it isn’t on par with satellites, but the modular phone may prove more difficult. Project Ara relies on lots of manufacturers in lots of different places to make its components in parallel without being able to test the things they’re making against other those things. It sound crazy! So Google had to develop a way to test not just every module, but how every module interacts with every other kind of module.


The only thing Google will make is the endoskeleton frame—the bones that you can snap all the other modules on to. It’s got frames for three sizes, mini medium and jumbo. Because the modules are interchangeable, you could conceivably have a few different size phones in your pocket. The mini looks sort of like an old iPod nano. It’s long and thin and just over two inches wide.


“I don’t understand why phones are getting bigger,” Eremenko says, flipping a mini frame around in his fingers. “Everything else is getting smaller. I’m going to buck the trend, I want to bring the mini to market.”


I argue that the big phones are very popular, and even better, are awesome. But Eremenko replies by noting that that’s the magic of this thing: you don’t have to decide. “As a consumer you could own multiple frames. You can have a mini to put in your skinny jeans and go clubbing, and a big one to take into the office and do email.”


It is an amazing vision. Amazing! Can you imagine taking the same phone clubbing, and then into work the next day where you use it to test urine specimens? What an age we live in.


So Google will build the frame, and it will also build a framework to help developers solve those pesky compatibility problems. It has a module development kit, or MDK that will let manufacturers test and prototype their components purely in software, without having to run them against all the possible hardware permutations. The idea is that if you confirm to the MDK, everything will work together, no matter what you throw at it. All you have to do is plug it in. It’s far-fetched, but there’s a precedent for it.


“The plug and play model has been solved in other platforms, just not in mobile,” says Eremenko. And then he points out all kinds of other examples that use a uniform data or power bus across the device that lets people add in parts. You can buy any number of different peripherals that plug into your computer’s USB port and will work just fine. Modern cars run on vehicle bus protocols that make sure all the electronics can talk to each other. Even airplanes use spacewire. But modern phones are different. Just about everything other than the battery and the SIM tends to be hard-wired into the device. “Mobile phones are just about the only thing that doesn’t have a bus,” he argues. “That’s a historical outlier we’re trying to fix.”


And they need to fix it soon, because part of the deal with ATAP projects is that they are time-limited to two years, which puts Ara on a collision course with the marketplace next Spring.


To get it all rolling, Google is hosting a second set of Project Ara developers’ conferences. This time, it’s taking the show on the road. There will be a conference in Mountain View on January 15 2015, followed by another in Singapore on January 21. It’s also got satellite sites in NYC, Buenos Aires, and London, and in Bangalore, Tokyo, Taipei, and Shanghai where developers can get together in Google’s offices to coincide with the conferences.


The reason they are all over the world this time—instead of just in Silicon Valley like the one in April—is that Google knows it has to get the world onboard. And not just the 6 billion people it has very publicly said it wants to sell one of these phones to, but the developers who will give those 6 billion people some sort of distinct reason to buy. A lot of those developers are international. Google needs the world to pile on its bus.


So here comes the developers’ conference, and with it a new prototype: the long-awaited Spiral 2. Spiral 2 will have custom-made chips instead of ones that are essentially emulators, like in the first device. The electro-permanent magnetic couplings will be smoother, so everything fits together in a more invisible fashion. And it’s got a design and manufacturing partner now, Quanta, out of Taiwan. All the earlier prototypes, like the one I’m here to see today, were cooked up by Newburg and his wife Ara Knaian (the project is named for her) in their Boston lab.


Project Ara engineer Seth Newburg holds a black box containing the most recent Spiral 1 prototype.

Project Ara engineer Seth Newburg holds a black box containing the most recent Spiral 1 prototype. Talia Herman/WIRED



And the hope, the real hope, is that Ara will attract all kinds of new developers, ones who have never been on mobile before. Sure, it needs the big tier-one suppliers who can make, say, a name-brand lens. And it wants to bring in manufacturers that haven’t gotten into mobile because volumes have been too high and margins too low. But the real energy is around the startups.

“One that came to me had a microfluidic sensor” says Eremenko. This sensor could test all sorts of fluids–blood, water, saliva, urine, you name it. Delicious! But while the company that made it is great at fluid-testing sensors, it isn’t very good at that other stuff it needs to go with the sensor—things like the industrial design or the real-time operating system that tells people what they’re looking at. “Ara solves all of that—not the least of which is: what is my retail channel.”


I mean, that’s if it works! Eremenko didn’t forget about the phone not booting. Nor did Newburg. Nor did I, after having previously seen it not boot in April and then, again, not boot in June and wondered if now, today, it actually would.


And without a lot of fanfare, Newburg connects it to and external power supply.


And he hits a button.


And the screen fires to life. Suddenly, there is the very familiar Android homescreen, with very familiar Android icons.


“This is a first in mobile! This is a device running over an internal network, rather than just everything being connected to a CPU,” boasts a now clearly excited Eremenko. And it is exciting!


Newburg taps an icon and a Stephen Colbert video starts playing. Next he dives into an app for controlling the electro permanent magnets. It looks like a traced off version of the device, with an outline of each app. He touches one of the module outlines—it correlates to a pulse oximeter on the device, which is basically an LED light—and the outline goes dark. He removes that piece of the phone, hands it to me, and swaps another in, but the phone stays powered on the entire time. I turn the little light over in my hand, and am, very truly, in awe.



No comments:

Post a Comment