Research On Artificial Nose-lung System And Odor Identification Algorithm Of Human-head Portraitrobot

With the increasingly complex of living environment, the identification and detection of mixed gas has become an urgent problem waiting for being solved. As a part of humanoid robot, the research on humanoid head portrait robot is mainly focused on the aspects of facial expression, speech, vision, and so on,and less on the robot’s sense of smell. Besides, although the correct rate of gas recognition in the current study of olfactory research is also relatively ideal, but the types of experimental gas species is less. The research of this article is that adds artificial nose- lung system to the experimental platform that is an humanoid head robot called H&F robot III. This paper takes five kinds of common and toxic gas: alcohol/hydrogen sulfide/ ammonia/smogand/ methane four kinds of mixed gas which includes hydrogen sulfide, smoke and methane gas as olfactory system’s objects. And this paper realizes the coordination of smell recognition results and the robot facial expression in the end.This paper designs and developments of artificial nose- lung system software and hardware aiming at human-head portrait robot. Artificial nose includes signal acquisition circuit and test container. Artificial lung is realized by the micro vacuum pump suction, expiratory function. Based on the hardware, the software and the host computer software are designed to realize the function of signal acquisition and transmission.This paper makes a comparative study of gas identification algorithm and designs program. Firstly, this paper uses K- clustering method to classify the gas. Then in this paper, the combination of genetic algorithm and BP neural network is used for identifying 5 kinds of single gas. According to the nine kinds of gas(five single gas, four mixed gas), this paper designs the pattern recognition algorithm based on three cascade neural networks at last.This article carries out gas identification experiments. Experimental results show that the correct rate of gas identification of five kinds of single gas is above 90%. And the correct rate of gas recognition of nine mixed gases in addition to hydrogen sulfide smoke gas mixture is 70%. The recognition accuracy of the other eight gases is above 90%. The experimental result shows that the gas identification algorithm is effective.The artificial nose-lung system is integrated into humanoid head robot H&F robotIII. It completes the coordination of output the recognition results and the experimental gas humanoid head robot expression at last. After repeated experiments, it can be saw that the robot can achieve similar facial expressions with the same odor stimuli.The research of this paper will make the humanoid-head portrait robot H&F robot- III closer and much more accessible communication with people.