IDF - final day. Bring on the robots!

Last keynote, and Justin Rattner is out and about on stage strutting his funky future. We have robots that follow an apple waved in front of them with the eager attention of a puppy tracking a sausage. We have programmable matter, which will know how to snap between a device large enough to type on and a small ball you can stick in your pocket - or a bracelet. We have power transmitted through the air. We have headsets that can respond to your thoughts and control virtual reality.

Or do we? The robot is real enough - it's demonstrating electric field pretouch, which was also on display at the last IDF: the fingers of the robot hand set up an electrostatic field and monitor how objects disturb it. That way, it can tell when it's coming close to something, which way that something is moving, and even what sort of something it might be, all without actually touching it. What was new this time was the translation software which moved the robots hand and fingers to track, hold, move and release - and it did so with something of the air of a cobra about to strike. We are deep in the uncanny valley.

The rest all had healthy servings of yes-but. The head-mounted thought detector - due out by Christmas as a games controller - is actually far better at detecting the signals from facial muscles and the angle of the head. Although the demonstrators claimed it could also work by thought alone, the demonstration was far from convincing and, talking to the chaps from emotiv.com - who designed it - afterwards, I didn't get any good technical explanations about how it copes with the hard problems of that particular task.

The programmable matter - or catoms - is still a long way from happening, although the lead researcher said that when he started on the project four years ago, he'd put it fifty years into the future. Now he thinks it's only a couple of years away. Certainly, the main new thing on display at the keynote - although apparently very dull - is in fact very significant. It was an array of tiny glass beads each a tenth of a millimetre across: the significance is that these are perfectly spherical and produced via photolithography, making them ideal for the construction of programmable matter in ordinary chip fab plants. Integrate electrostatic alignment and motility, tiny power sources, sensors and processor smarts, and you're there.

And finally, we get to the wireless power transmission. This is the stuff of pre-war horror movies - giant copper coils, banks of glowing test equipment and a light bulb that fades in and out in an eerie fashion despite not being connected to any form of visible power. I first came across this idea when MIT researchers reported on it, and had a hard time understanding what was going on. This time, I collared the unfortunate researcher who had been chosen to go on stage and got to ask questions. Lots and lots of questions.

And the secret is - it's nothing new, at least not on the surface. The principle has been used in just about every radio since the 1930s: you have two coils tuned to the same frequency; put power into one at that frequency and power comes out of the other. Nobody's used it to couple much power before though: rather, it's been used to clean signals of frequencies away from the resonance.

There are lots and lots of things to sort out before the process, which Intel calls WREL for Wireless Resonant Energy Link, becomes useful. Even if everything is perfected, it's unlikely to work over more than around ten metres (although that may be more useful than you think). But it looks awesome, and some of the applications in conjunction with other new technology are particularly intriguing.

And then it was all over, with nothing left to do except meet Steve Wozniak and catch the plane back home. More to come, though, when I'm safely back in Blightly...