Intel CTO demos building blocks of shapeshifting robots
When the end comes, it'll be wearing a bunny suit
IDF Intel sent sane journalists screaming for the exits this morning when it unveiled a nightmarish future vision where robots are more intelligent than humans, computers can change shape, electronic devices are recharged remotely, and humans are probably going to be ruled by an x86-based server farm.
Wrapping up the Intel Developer Forum with a session on R&D, CTO Justin Rattner declared that the singularity – the point at which machines become more intelligent than humans – is now as little as 50 years away.
For some reason, the MSc and PhD-heavy attendees did not see this as a horrifying prospect and run for the hills, but instead sat there dumbly while the vendor recounted the ways it is underpinning the necessary research to ensure machines can outwit humans.
The presentation started sensibly enough, with an update on new materials to take microprocessors beyond CMOS and 32nm, and subsequently further jack up computer power. These inevitably include carbon nanotubes and other carbon-based materials and emerging materials program manager Mike Garner raised the prospect of new materials which use spin, magnetics and optics to encode data, instead of charge, and which are able to offer more than one state.
The vendor also wheeled out its usual update on silicon photonics, with a researcher demonstrating a laser based system where data was being modulated at 2.2Gbps. He said by tuning lasers at different wavelengths and integrating them together a 1Tbps link could be possible.
Another IDF favourite, software configurable radios was also highlighted, with a call for wireless networks to be rethought from the ground up, so that devices could sniff out the most effective spectrum, rather than being restricted to a single chunk by the FCC and other regulators.
Then it started getting weird, before getting just plain scary.
Intel researchers demonstrated a Wireless Resonant Energy Link. Think of two antennas, one hooked to a a power source, the other hooked to a nothing else but a lightbulb. By using resonant coupling, power was transmitted from the power source to the light bulb, to the tune of 60W at 75 per cent efficiency. Or put simply, the lightbulb lit up, purely on wirelessly transmitted power.
Rattner said the firm envisioned receiving resonators being built into laptops and mobile phones, so that they could be recharged quickly simply by being placed next to a power transmitter. He added that the technology was – probably – perfectly safe, as it relied on resonant coupling, not anything nasty like inductive coupling.
He then wheeled out pair of robots, to demonstrate Intel-backed academics’ efforts. One robot was trained to pick out mugs amongst clutter, and grasp them in a safe manner. Another used “electric field pre-touch” rather than a camera to suss out the world and grasp objects This is a technique used by animals like sharks, and sure enough the predatory robot quickly made a grab for the Intel VP, shouting, “found it, got it, found it got it.”
Non-touch manipulation was the party trick for San Francisco-based neural interface firm Emotiv, which demonstrated a headset which detects users’ brain waves at the skull surface to manipulate objects in a computer game. This should be on the market for Christmas the firm said. While the headset would indeed add new element to Star Wars games, the truly paranoid will be asking how long it will be before the machines start sending information back the other way.
And if Rattner’s last party piece was anything to go by, we won’t even see the machines coming.
Rattner treated the still unscared audience to a demo of Intel’s progress in programmable matter, or the use of millions of tiny micro-robots – dubbed Catoms – to build objects that can change shape. That’s right, shape shifting robots – we’ve seen those in movies.
Intel-backed researchers have built chunky prototypes, packing a microprocessor and memory, which are encased in multiple electromagnets which can then “roll around one another’s perimeter. It has gotten prototype cylinders down to 2mm, at which point electromagnets are too clunky, and electrical fields are used to manipulate the catoms.
The next step is 3D spherical catoms, at 0.1mm, and researchers showed Intel wafers with tiny silicon hemispheres, which could eventually provide the basis for fully-fledged catoms.
Catoms raised the possibility of 3D visualisation, for example in medicine or product development, Ratter said, or on the every day level devices that could morph from being a tablet computer to a pocket device.
Or, of course, a killer robot, controlled by a distributed neural network which is powered by electricity given off by its farm of human power drones. We’ll know we’re in trouble when the machines decide to rip off their Intel Inside stickers.®