Friday, October 16, 2009

Mouse Plays Quake II or The future of Organic AI

From the realm of awesome... check out this MOUSE (hooked up via some sort of neuron brain linkage AKA a projector and a ball LOL) PLAYING QUAKE 2.  Brought to you by the awesome blog

Excerpt from the ORCON link:

Navy Declassifies Details
Of Pigeon Guidance Project

A study of missile guidance by pigeon pecking has been taken out from under wraps by the Navy. At the same time, perhaps to calm fears of guidance designers, the Navy made clear that the project has been discontinued.
Started during World War II, Project Orcon (for organic control) was a try-anything approach to solving some then-current problems. Guidance systems for homing missiles were being easily countermeasured and the Navy thought animals might have potential as a jam-proof control element. Pigeons were selected for trial because they were light, easily obtainable and adaptable. Their job was to ride inside a missile and peck at an image of a target picked up by a lens in the missile's nose. The pigeon's pecking of the target image was translated into an error signal that corrected the simulated missile's simulated flight.
The project was revived in 1948 and carried further. In simulated rocket tests, the pigeons produced "surprisingly good results." The researchers were convinced that a pigeon could successfully guide a speeding missile under optimum conditions, compensating for his own and the missile's errors.
But after three years of equipment development and testing, the project was abandoned because the range of the Orcon system could be no greater than the range of any optical system and the system could be used only in the daytime. The trainer used target images photographed in color by a jet plane, which made picture-taking dives at a destroyer and a freighter in open sea.

About the range limit - they should totally hook the pigeon up to a long range infrared view system.

So now I extrapolate... we should really get Mice to man the Cameras on the Border OR as my friend Ian M. theorizes... perhaps we should use mice as AI for virtual border crossing simulations.

No comments: