Study On Odor Source Localization System Based On Bionic Olfactory

The bionic olfactory system is also called "electronic nose". It has extensive application inmany fields, such as defense, military, industry, environmental protection, health, safety, etc.The traditional electronic nose can extract some meanful information (type, concentration andcomponents) of gas plume based on the response of its gas sensor array. But these functionscannot satisfy people’s demand for electronic nose. In recent years, some researchers are tryingtheir best to study how to utilize electronic nose to determine the direction of gas plume andhow to search for the odor source location by a mobile robot equipped with electronic nose.This is also the main work of this paper. Existing method designs an odor compass equippedwith distributed gas sensors, and make use of it to determine the direction of gas plume byanalyzing the difference of sensors’ response signal. However, this method faces the problemthat how to reduce the measuring error of gas plume’s deflection angle.Based on the existing gas sensor technology and the localization principle of odor compass,a new odor source localization system, which is similar to mammalian nose, is designed in thispaper. The system has two mutually perpendicular airways and each of them has a gas sensor onthe bottom. The system can rotate freely within the scope of360degree in horizontal.Compared with an existing odor compass, the gas plume information acquisition method of thesystem is different. The existing odor compass generally collectes gas plume information at astable angle, so the angle range is limited and the error is relatively large. The system in thispaper can continuously collecte gas plume information in the process of rotating to jugde theodor source direction by imitating human or animal keep running noses. This can effectivelyimprove the accuracy of the odor source localization.The gas sensor is TGS822, which is metal oxide sensor with a stable performance. Thecore of the sampling system is the MCU MSP430F5529with a high performance and lowpower, which improves the compatibility between sensors and sampling systems. So it can wellmeet the requirement of odor source localization experiments. The angle between gas plumeand one airway is set as deflection angle, and the experiments collect every5degree from0degree to30degree’s gas plume data. The feature extraction method combines K-L transformwith Fisher discriminant. At last, in order to verify the feasibility of the system and scheme, the classification results are tested by test sample set.Experimental results show that the angles’ classification works well under laboratoryconditions by using combine K-L transform and Fisher discriminant method to extract feature.In other word, even if the deflection angle is as small as5degree, it is also possible todistinguish the gas plume angle. The angle discrimination of the exsisting odor compass for gasplume is30degree, so the odor source localization system for gas plume angle localization ismore accurate. It has a great significance for the application which demands for high angleaccuracy.