Performance Evaluation of PSO and its Variants for Odor Source Localization

Abstract

One of the most intriguing areas in robotic applications is odor source localization. Particle swarm optimization (PSO), one of many nature-inspired algorithms, has been preferred by many researchers due to its simplicity ease of use, and good performance for odor source localization. However major challenge with PSO is handling local optimum and maintaining diversity among robots. This paper aims to provide a systematic review of PSO-based methods for odor source localization, and a performance comparison of standard PSO and its several variants in two distinct environments. To evaluate the effects of the intermittent nature of odor plume on the performance of odor source localization methods, odor plume distribution is modeled using the Gaussian model and Ansys fluent in environments 1 and 2 respectively. Parameters, average time consumption, the average number of steps needed, and average success rate are measured to compare the performance of PSO and its variants for odor source localization.