Research Of Robot Nose And Odor Source Localization

More and more attention has been paid to the research of robot nose since 1990's. Odor source localization is an important subject in the research of Robot Nose. A Lot of researchers in foreign counties have devoted themselves to the area, but few have been seen in China.In the dissertation, the design of robot nose system including software and hardware is introduced, two signal processing methods for infrared gas sensing are analyzed, and a new odor source localization algorithm is proposed and validated in the simulation and experiment.The sensor of robot nose is the key part of the system. The metal oxide gas sensor has high sensitivity and its measurement circuit is very simple, but it has the drawback of long response/recovery time. The infrared sensor has the advantage of short response and recovery time, but the measurement system is complex. Our robot nose contains an infrared gas sensor and four Figaro metal oxide gas sensors and features the realtime measurement ability. The infrared gas sensor contains an open-path gas chamber, a high SNR (signal to noise ratio) infrared detector, and an electrical modulated MEMS infrared emitter. Two signal processing methods, one based on analog filters, the other based on digital lock-in amplifying technique, are developed and analyzed. Each method helps the system achieve high sensitivity and high SNR.The system hardware has been designed based on TI's TMS320F2812 32bit fixed-point DSP that enhances the realtime processing ability. The system hardwareis divided into three units. The DSP unit performs the computing task. The Analog unit performs four tasks, the sensor signal conditioning, AD/DA, modulation of the infrared emitter and control of the heating voltage for Figaro sensors.The system software has been designed based on the embedded operation system, μCOS-II. The software is divided into several tasks with different priorities to meet the demand that the system could process the sensor data, implement the odor source localization algorithm, and provide the control logic for robot car. With the help of uCOS-II, not only the real-time performance is improved, but also the process is much clear and it is easy to add more functions.A new algorithm for odor source localization, Simplex Based Grid Searching Algorithm, is proposed and has been validated in the simulation and experiment. It treats odor source localization as a function optimization task. It makes the best of the concentration grads information, fuses the data of sensors, and allows relatively low control precision. The system simulation has been performed based on simplified models, and the result shows that the Simplex Based Grid Searching Algorithm has higher probability to find the odor source than the algorithm of instantaneous concentration grads searching. The experiment has been implemented in weak wind environment, and the result shows that the new algorithm is practicable.