Fuzzy hierarchical pattern recognition for robotics applications

Abstract

Two approaches of pattern recognition for robotics applications are introduced. The first study is concerned with a method of efficient pattern classification for moving objects using a discriminant tree. The second study is about three dimensional pattern classifications. Both studies use fuzzy logic and hierarchical knowledge base. In the first study, the experimental system shows the robot-arm system which is able to recognize a moving pattern (parts shape recognition) and to manipulate a moving object on a belt-conveyer at a various speed. And in the second study, the experimental system provideds a car-name recognition system which is able to recognize various car-name (trade name) from image data of minicars.