Localization of remote odor sources by metal-oxide gas sensors in turbulent plumes

Numerous applications reported in the sensors literature have established metal-oxide gas sensors as effective devices for detecting and quantifying a broad range of chemical events. A critical question that could substantially expand their usage is the following: can they characterize the spatio-temporal stimulus characteristics for predicting the source location? We show that commercially-available metal-oxide sensors have a high enough temporal resolution for such a characterization. By selecting proper features of the sensor response and by maintaining a representative response database recorded from known locations in similar plumes, simple short-time measurements can accurately predict the displacement from the source to the sensing location. A key advantage of such a prediction scheme is its achieving the localization remotely and without a navigation within the plume.