Learning from the Moth: A Comparative Study of Robot-Based Odor Source Localization Strategies

Abstract

The odor search strategies of the moth have been researched since many decades. Many behavioral studies have described the behavior under well controlled conditions, making predictions on what the underlying mechanisms might be. However, it is almost impossible to asses these mechanisms directly since sensory and behavioral data on a freely behaving moth are very hard to obtain. Therefore, we propose a comparative study were the behavior of a robot is analyzed when controlled by a number of odor source localization models. Our results show that a system making use of stereo odor information outperforms some well‐established chemical search models.