The Universal Planar Manipulator

macsox writes: "Wired News has an article about surfaces that, using vibrations, can move objects around at the owner's whim -- for example, using a mouse as a remote desktop arranger. Also envisioned: rooms that redecorate themselves. The scientist's page is here."

Make sure this doesn't fall into the wrong hands..

..or we'll all be running around in the streets, being chased by furniture, screaming, "HELP!! CHAIR!! HELP!! CHAIR!!"

Vibrating Magic Remover

Dan Reznik, a Brazilian computer science Ph.D. candidate at the University of California at Berkeley, has developed a table with a shaking top that moves objects as if by magic -- the only finger lifted is the one on the mouse controlling what objects are moved where.

Combine this effect with the pressure-sensitive polymer from a few days ago and some control electronics, and you'd have a floor able to react automatically to the presence of a life insurance salesman by wisking him effortlessly back out the door.

gamers take note

and what if we put on the table 22 little plastic men, 11 each from two opposing football squads... oh yeah.

Neural Net Vibration Control

Actually there is an entire field of neural network vibration control that started out as adaptive vibration cancellation [unh.edu]. In a 1989 demonstration for the International Joint Conference on Neural Networks, SAIC [saic.com] had a demonstration where they placed 1, then 2 then 3 accellerometers at various arbitrary places on a 3D grid structure being stimulated by some vibrators similar to those spoken of in the article here. Then the outputs of those accellerometers were fed as "pain" signals to a recurrent neural network that controlled some other vibrators. When the neural network was turned on, the vibrators under its control would vary frequency, phase and amplitude until vibration was cancelled out at precisely the 3 locations at which the accellerometers were placed. You could then pick the accellerometers up and put them back down somewhere else and the neural network would adapt within a few seconds, cancelling out its painful inputs.