4 new Google projects: a couple of words about Soli, Jacquard, Vault and Abacus


At the recent Google I/O conference, the ‘corporation of good’ announced (in addition to new Android versions, new services, and other predictable announcements), four new projects: Soli, Jacquard, Vault, and Abacus. Let’s see what these curious techs are about and how they can make our world a better place.

Project Soli is a tiny sensor designed for smart watches or fitness bands to more accurately track hand movements and enable better control by gestures. The gestures would be used both for device control (a more convenient option than trying to manipulate tiny objects on an equally tiny screen) and ecosystem control using a wireless connection as Bluetooth to transmit commands.

As we know, today’s gesture recognition technologies employ, mostly, the principle of analyzing an image recorded by one or more cameras. But it is a costly system as it requires a number of components, and it is too demanding in terms of computing power and energy consumption.

The Soli sensor is, in essence, a minuscule radar operating on the 60 GHz frequency using just two antennas. The sensor scans 10 ‘frames’ per second, and the reaction is practically instantaneous (which is a priority in any similar interface).

Besides that, you don’t need to wave your hands like mad as with Kinect: Soli is capable of detecting micro-gestures like rubbing your fingers or snapping.

It’s not entirely clear how ‘fast-learning’ this interface would be: It is quite obvious that, for instance, multi-touch gestures, excluding the most basic ones, are used by, literally, no one. The most critical task would be to ensure finger gestures are extremely intuitive and predictable, and not require a user to sustain long drills learning and exercising the gestures.

Project Jacquard (noun; a type of cloth) is designed to dramatically transform the fashion industry. The idea is based on integrating multi-touch sensors into the cloth, so we could use a spot on our sleeve or lap instead of a smartphone screen.

Besides the sensor surfaces themselves, the developers of the project Jacquard took care of special colored conducting threads to connect sensors to controllers – the threads would be clandestine and would not screw up the outfit design. Or, if you wish, they would be visible to let the people around know you are wearing something unconventional, trendy, and hi-tech.

It is critical to note that, from a technological perspective, the principle of integrating sensors into cloths has already been elaborated in a fashion that does not raise the costs of producing such cutting-edge outfits – all you need is the conductive threads. It is not a concept bound to come true 146 years from now: Google has already boasted its agreement with Levi’s, so the cybernetic jeans are expected to arrive soon.

The use cases for them which immediately spring to mind are, of course, an ability to input data without taking out any gadget with a touchscreen (for instance, to dial a phone number) and to serve as a substitution for gaming controllers and other conventional means of input.

Project Vault is a brand new technology for protecting data against unsolicited access, which is forecasted to be even more secure than a fingerprint scan. Its meta-objective is to finally get rid of passwords. Easy passwords, which are memorized due to their simplicity and sensibility, are also easily cracked by today’s computers in mere minutes. Complex passwords (the likes of Xj$7f(sQp]1v^4), on the other hand, cannot be memorized by anyone, so it ends up being written on a Post-It note, which is nowhere near being secure.

Vault is a very compact device with a form factor of a microSD card. “Real’ computers or smartphones, regardless of the OS it operates, detect the card as a detachable storage device containing two files: one to write into and one to read from.

In fact, it is not a simple memory card. With all the hardcore tech jibbering aside, it is a virtual computer, which is completely autonomous and has the job of encrypting and decrypting data, whereas external software, whichever OS it is made for, performs a checksum to make sure the file read matched the file written and the key is accepted.

You can gain access to important data from any device, but only if you happen to possess this ‘wonderflash-drive’ containing, besides non-volatile memory, an ARM processor, an NFC module, and an antenna, all wrapped in a single solution managed by the Linux-based RTOS operational system.

That means, for anywhere-any time authorization purposes, one would be required to simply plug in the storage and not put in a password. The whole idea is not revolutionary: All in all, it is just the same old USB token or its predecessor, an LPT key, which are used by many in online banking applications. The drawback of those traditional solutions is the inability to get access from other devices: A user has to have previously prepared settings and install drivers to make both work on a PC.

During the demo, the Vault project was showcased at the secure messaging demonstration. An outbound message encrypted on one smartphone, is then transmitted over open-protocol networks and is decrypted on the destination device. No keys are passed, thus they cannot be intercepted.

Project Abacus is a completely new approach to user authentication. Instead of using a password, or two-factor model, which relies, again, on a password and some other authentication method, Google’s researchers propose real multi-factor authentication, based on your location patterns, how you talk, how you type, etc. All these factors combined allow identification of a concrete person much more reliably than a password.

The most amazing news here is that this technology doesn’t require any specific hardware. Everything project Abacus needs to operate already exists in every modern smartphone. It’s just about adding the software. On the other side, this approach requires the gathering of a lot of information about user behavior, which is quite uncomfortable, taking into account how much of our data Google already has.