Performance Analysis of PSO‑Based Path Planning Using Chaotic Systems with Different Chaoticity Levels

Abstract

In this study, the performance of a particle swarm optimization (PSO)-based meta-heuristic path planning algorithm integrated with different chaotic systems was evaluated. Unlike previous studies in the literature, this research specifically investigated the relationship between the chaoticity of chaotic systems and path planning performance. Lorenz, modified-Chameleon, and scaled-Zhongtang chaotic systems were integrated into the PSO algorithm by replacing the random parameter values required during each iteration with chaotic signals generated from these systems. The chaotic characteristics of the selected systems were analyzed using phase portraits, Lyapunov exponent spectra, and fast Fourier transform (FFT) analysis to determine their chaotic behavior levels. To evaluate the proposed approaches, seven different 20 × 20 m. test environments containing circular obstacles with varying positions and sizes were generated. Each algorithm was independently tested 30 times in every environment, resulting in a total of 840 experiments. The obtained results demonstrated that the scaled-Zhongtang-based PSO algorithm achieved the shortest path lengths and the best overall path planning performance in all environments, whereas the conventional PSO algorithm produced the lowest performance. Considering both the chaoticity analyses and the path planning results, it was observed that systems exhibiting stronger chaotic behavior provided greater improvements in PSO-based path planning performance.