A Comparison of Different Searching Strategies to Locate Sources of Odor in Turbulent Flows

Abstract

Four different searching strategies to locate the source of odor in turbulent flows have been computationally tested. These algorithms can be separated into two classes. One is based on the rate of odor patches encountered by the searcher, the other employs the variation of odor concentrations as well as the rate of odor patches experienced by the searcher. The concentration of odor has been simulated by using a stochastic model for the time-evolution of concentrations along the path of a moving observer in an inhomogeneous plume. For each algorithm, we released the searchers at a crosswind location away from the plume centerline; we then calculated the ensemble average position relative to the plume centerline and the distribution of the searchers along the crosswind direction at the source. Compared with strategies using rate of odor patches only, the algorithms that also employ the variation of concentrations are seen to be more effective in locating the source; that is, the average path of all searchers is more biased toward the plume centerline and their crosswind distribution is more skewed toward the source. The findings of this work can be used as guidelines to discriminate successful strategies that can be tested in future experiments.