Multiagent visual area coverage using a new genetic algorithm selection scheme

Using genetic algorithms (GA) for solving NP-hard problems is becoming more and more frequent. This paper presents a use of GA with a new selection approach called the queen GA. The main idea is not to select both parents from the entire population, but to create a subgroup of better solutions (the queen cohort), and to use at least one of its members in each performed crossover. We demonstrate the use of the queen GA for the problem of repositioning observers across a polygonal area with obstacles in order to maximize the visual area coverage for a given time horizon. The queen GA gives superior results over a GA with different selection methods (i.e. proportion, ranking and tournament) at the 0.01 significance level. These comparative results were duplicated when elitism was included.