Chemical Source Localization Using Electronic Nose Robotics


Experimental Overview
Experimental Setup
Data Acquisition
Data Processing
Discussion and Future
About Us


According to recent biology literature, Mustelus canis sharks localize odors based on the time delay between olfactory signals detected by each of their nares.  This project replicates this process by implementing a LabVIEW GUI to collect voltage data from an array of three sensor pairs, in which each sensor in a pair functions as an individual nare in an “electronic nose.”  This project investigates several methods of isopropyl alcohol exposure to simulate the dispersion of odor due to a chemical leak.  Experimenters implemented a Dynamic-Time-Warp processing algorithm to compute the overall delay between signals in each sensor pair. These delays were used in a voting system to localize the source relative to the sensor array.  This approach presents some limitations because signals from individual sensors in each sensor pair do not have identical morphologies.  While additional work is required to improve the system’s consistency, preliminary results are promising.  This sensor system will eventually interface to a mobile robotic platform that will actively seek out a chemical source.


Electronic sensing technology is a developing field of study that has greatly advanced over the last decade in technical and consumer applications. Electronic noses are already being introduced in research laboratories, manufacturing processing technology, home and workplace safety monitoring and quality control. We initially attempted using a concentration gradient to locate the source of an odor, but after puzzling results and the publication of  "The Function of Bilateral Odor Arrival Time Differences in Olfactory Orientation of Sharks," by Jayne M. Gardiner and Jelle Atema, in Current Biology this summer. Prior to the release of this article, animals were believed to orient themselves towards an odor by comparing bilateral odor concentration differences, turning toward higher concentrations. However, in the article, a small shark species, Mustelus canis, was presented with with carefully timed and measured odor pulses directly into their nares. They turned toward the side stimulated first, even with delayed pulses of higher concentration. This is the first conclusive evidence that under semi-natural conditions and without training, bilateral time differences trump odor concentration differences. This project applies Gariner’s and Aetma’s findings to the development of an artificial sensing system to localize an isopropyl alcohol source.  The goal was to enable this sensing system to correctly choose between two possible chemical source locations (referenced as ABOVE and BELOW the sensor array in this report) based on differences in the time of activation of individual sensors within the system, just as Mustilanus canis sharks choose one of two possible directions towards an odor source primarily based on bilateral time differences.


This project used a method known as Dynamic Time Warping to determine which sensors within the system were activated first.  Dynamic Time Warp algorithms are ideal to compare two signals that vary in time or speed.  For example, they are often used for applications such as speech recognition or gait analysis to account for variations in speaking and walking speeds1.  The algorithm is particularly useful for this project because, while sensors further away from odor sources are expected to respond later than those closer to the source, the actual time delay between these responses is highly variable due to random air currents.  Dynamic Time Warping allows the signals to be warped in the time dimension in order to create a best possible match that is independent of variations in time or speed, so that it is easier to determine which signal was produced first.



Potential Applictaions:

The designed GUI and experimental setup can be used as a starting point for future research exploring chemical array signal processing applications, such as chemical source localization and  specific compound detection.