Investigation on selection schemes and population sizes for genetic algorithm in unmanned aerial vehicle path planning

Abstract

Genetic algorithm has been widely used in numerous fields. Recently, researchers have developed significant interest in applying GA in unmanned aerial vehicle path planning. However, experiments should be conducted to determine selection schemes and population sizes. This process is time consuming and inconvenient for users. Thus, a study is conducted on the performance of GA at various selection schemes and population sizes. Results show that large population sizes do not contribute in improving the performance of GA. Tournament selection shows the highest performance in terms of path cost, whereas the weakest performance is in terms of computational cost. Truncation selection shows the optimum performance among the selection schemes.