Odour source localization strategy for multiple robots using swarm intelligence with odour-gated anemotaxis

Abstract

Animals such as silkworm moths, dogs and blue crabs have exhibited odour localization capabilities in nature. This amazing ability is exhibited in complex airflow conditions which produces highly dynamic and unpredictable gas dispersion. Harnessing this capability will enable robots to be deployed in critical and high-value applications such as search and rescue, entry point security applications, and environmental monitoring in industrial and urban settings. This thesis documents the research in swarm intelligence for gas source localization. Using swarm intelligence to achieve this task is envisaged to be more practical and economical compared to single robot implementations. Currently, few works have been presented on multi-robot systems in gas source localization; much less using swarm intelligence. This research aims to fill in the research gaps in gas source localization using swarm intelligence. However, current mobile olfaction experimental methods tend to oversimplify the actual problems thus reducing the findings’ impact in advancing this research field. Furthermore, lack of experimental realism reduces the reproducibility of the previously presented works in real world conditions. To overcome this limitation, this research uses recorded real-time gas dispersion in robot simulations. A real-time gas dispersion monitoring system consisting 72-gas sensors was built to record the gas dispersion in the experiment area. The recorded real-time data stream was then used in simulations to accurately recreate realworld experimental conditions in a simulation environment.