Fuzzy relations for behaviour-fusion of mobile robots

A "sophisticated" behaviour of an autonomous mobile robot is composed of several "primitive" behaviours. This is achieved using the so called "behaviour-networks". These circuits are built out of fuzzy gates defined to perform binary and unary operations. Two classes of binary operations are defined over the set S=X/spl times/Y, x/spl isin/[x/sub max/,-x/sub max/], y/spl isin/[y/sub max/,-y/sub max/], x,y/spl isin/R. The first class contains symmetric operations and is similar, in a certain sense, to the logical OR, AND and XOR. The second class includes asymmetric prevalence operations. Unary operations such as clipping and extending are also defined for conditioning the signal prior to the application of the binary operations. Although the main motivation for the development of these "fuzzy gates" was behaviour-fusion of mobile robots, they can be used as universal circuit elements for fuzzy-logic operations. An example shows a mobile robot with several primitive behaviours trying to reach a goal position whilst avoiding collision with an obstacle.<>