Multi Agent Visual Area Coverage Strategies using an Adaptive Queen Genetic Algorithm

H. Stern, Y. Chassidim, M. Zofi, and M. Kaspi (Israel)


Genetic Algorithms, Adaptive, Covering Problems, Visual Search, Visual Area Coverage, Multi Agent.


Using Genetic Algorithms (GA) for solving NP-hard problems is becoming more and more frequent. In an earlier paper we proposed a GA with a new selection approach, where one parent is selected from the entire population and the other parent is selected from a sub group called the queen, containing the K best individuals. We referred to this version as the queen genetic algorithm (QGA) In this paper we present an improved QGA, called the adaptive queen genetic algorithm (AQGA) The improvement includes an adaptive queen size and mutation rate, allowing both to change over the generations according to the population's diversity. We also present a generalization to three dimensions of the visual area coverage problem (VAC). The VAC problem involves moving dynamic observers across a 3D rough terrain in order to maximize the visual area coverage for a given time horizon. We demonstrate the use of AQGA with different settings to solve the VAC problem, and compare the results to the QGA.

Important Links:

Go Back