Research On Target Odor Discrimination And Source Location Based On Odor-sensitive Mobile Robot

With the development of modern technology,machine /mobile robot olfactory plays an increasingly important role.Robots with sensitive olfactory perception can reduce casualties and losses in hazardous gas detection tasks.Among them,odor identification can obtain rich odor category information,has the advantages of anti-interference and high predictability,and should be placed in a high priority position in gas sensing tasks.Because of the limitation of traditional single-type odor source localization,based on multi-legged mobile robots,this thesis starts from the perspective of tracking dogs first identify and then track the dangerous goods,integrating target odor identification into the leak source location process,a target odor discrimination and source localization method based on odor-sensitive mobile robot is proposed.Through the effective combination of target odor identification model and spatial information odor concentration gradient,the target odor identification and source localization of gas sensitive mobile robot in indoor sampling environment are realized.The main works of this thesis are as follows:Firstly,the three gas sensing tasks of mobile robot olfactory are described,and the relevant research status at home and abroad is summarized.Then the importance of target odor identification in gas leakage is analyzed,the basis of target odor identification and source localization are summarized,and its principle and advantages and disadvantages are analyzed.Secondly,a sliding window-decision tree target odor identification model is proposed.According to the characteristics that the odor information of the gas-sensor cannot reach a steady state in the open sampling environment,on the basis of the decision tree classifier which based on the Gini index,the sliding window technology is introduced to extract the short-term sequence features from the original transient odor information to realize the identification of odor information at the current time.Thirdly,the search strategy of spatial information odor source by fusion target odor identification is proposed.In view of the problem that the gas sensors are prone to "drift" and the traditional odor source localization method cannot identify odor categories in the environment,this thesis integrates the target odor identification algorithm into the process of odor source localization for fusion detection to improve positioning efficiency and accuracy.Finally,the hardware module of gas sensor array is built and mounted on the CR-6multi-legged bionic robot.The olfactory analysis experiment scheme is designed to collect the odor information by the gas sensitive multi-legged mobile robot.The validity and accuracy of the proposed method are verified by semi-physical simulation experiments.