Abstract: The topographic information
of a closed world is expressed as a graph. The plural moving
objects which go and back in it according to a single moving
strategy are supposed. The moving strategy is expressed by
numerical values as a decision table. Coding is performed with
this table ad chromosomes, and this is optimized by using
genetic algorithm. These environments were realized on a
computer, and the simulation was carried out. As the result, the
learning of the method to act so that moving objects do not
obstruct mutually was recognized, and it was confirmed that
these methods are effective for optimizing moving strategy.
Key words:
Genetic algorithm Graph theory Strategy Cooperative behavior Machine learning
Manuscript received on
December 15, 1994
|