Comparison of Genetic Algorithms for Evacuation Route Optimization

Abstract

Efficient evacuation route planning is a critical task in complex multi-story buildings. This study presents a comparative analysis of four Genetic Algorithm (GA) variants—Standard GA (SGA), Island Model GA (IMGA), Hybrid GA-PSO, and Adaptive GA (AGA)—for solving the Shortest Path Evacuation Route Problem (SPERP). The building environment is modelled as a weighted graph G = (V, E), where vertices represent spatial components and edges represent traversable paths with associated costs. Simulation experiments are conducted using real architectural layouts of Gedung Depan and Gedung Belakang. Results indicate that Hybrid GA-PSO achieves the fastest convergence, the shortest average path length (40.1 m), and a 100% success rate, outperforming other approaches by 35–55%. Both the Adaptive GA and the Island Model GA demonstrate strong reliability through dynamic parameter control and population diversity. These findings confirm the effectiveness of hybrid and adaptive evolutionary algorithms for real-time evacuation optimization.