The creation of computers capable of making fuzzy matches recognising that when someone types teh or comapny they probably mean the and company, for example, has been a goal of designers for a couple of decades now, but Takeshi Yamakawa of Kumamoto University in Japan reckons that he may have made a breakthrough. According to Newsbytes, he has developed a microprocessor capable of drawing the most likely conclusion from fuzzy data. And, says Newsbytes, the chip has so impressed scientists of the US National Aeronautics & Space Administration that they have sought and been granted permission to incorporate the chip into the operation of forthcoming iterations of the Space Shuttle as part of the control function to help the Shuttle monitor and maintain its altitude in space and conduct docking manoeuvres with another space vehicle. Once NASA’s tests are complete, Yamakawa hopes to offer the chip on the commercial market. In order to be allowed to use the chip, NASA had to pledge to use the technology only for civilian space science and technology. The microprocessor is made up of a rule chip and a defuzzifier chip. The rule chip operates at 1m calculations a second to make inferences based on the fuzzy set theory first presented in 1965 by US mathematician Professor L A Zadeh. The defuzzifier chip converts the conclusion drawn by the rule chip into analogue numbers. Yamakawa last year built a fuzzy computer consisting of 10 boards, each with fuzzy inference circuits, and his new chip integrates the circuitry of the entire 10-board computer. The key to fuzzy applications is the use of analogue computing techniques where instead of being represented by strings of zeros and ones, the imprecise fuzzy information is represented by continous variables from zero to one.
