Ryosuke Konishi

The concentration-estimation of inflammable gases with a semiconductor gas sensor utilizing neural networks and fuzzy inference

This paper proposes a method to estimate the concentration of inflammable gases from transient response patterns which a semiconductor gas sensor shows under periodic heating conditions. The procedure and effectiveness of the method were illustrated for five selected gases of butane, hydrogen, LP gas, methane, and town gas. The response patterns obtained were found to be well reproducible and specific to the kinds of gases. Frequency analysis could be applied easily to the response patterns because of their periodic characteristics, allowing one to extract D.C. and A.C. components of them by fast Fourier transform. The A.C. components remained almost unchanged irrespective of the variations of ambient temperature and/or humidity and gas concentration, proving themselves to be adequate for the concentration-independent discrimination of gases. The D.C. components, on the other hand, depended largely on the variations of gas concentration, being useful for the estimation of gas concentration. It was shown that the discrimination of the five gases supported by a three-layered back propagation neural network as well as the estimation of their concentrations assisted by fuzzy inference were successfully performed.

The discrimination of many kinds of odor species using fuzzy reasoning and neural networks

Abstract To discriminate many kinds of odor species, a system composed of multiple gas sensors and neural networks is proposed. Three commercial gas sensors are used for the system, and four kinds of inflammable gases, four kinds of fragrant smells and one kind of offensive odor are introduced as odor species. The discrimination is performed in two steps to increase the efficiency of the system; the first step is classification of the odor group, that is, the groups of inflammable gases, fragrant smells and offensive odor; the second step is the discrimination of individual odor species in the classified group. 100% group classification rate is obtained by the use of simple fuzzy reasoning and the steady-state response patterns of the sensors. The discrimination of individual odor species is performed with a neural network and transient response patterns of the sensors and a high discrimination rate (99.2%) is achieved.

Analysis of efficient lip reading method for various languages

The traditional researches targeted at only one language, and there is no research to refer the language and recognition method. Moreover, a lot of model-based methods use only an external lip or intraoral region, and tooth or tongue region is not reflected to the feature. This paper describes analysis of efficient lip reading method for various languages. First, we applies active appearance model, and simultaneously extracts the external and internal lip contour. Then, the tooth and intraoral regions are detected. Various features from five regions are fed to the recognition process. We set four languages to be the recognition target, and recorded twenty words per each language. As the result, proposed trajectory feature based on three shape features, the area and aspect ratio of internal lip region, and area of intraoral region, was obtained the highest recognition rates of 93.6%, compared with the traditional methods and other regions.

Analysis of the sensing mechanism of tin dioxide thin film gas sensors using the change of work function in flammable gas atmosphere

Abstract To investigate the sensing mechanism of SnO 2 thin films prepared with the targets-facing type sputtering method, the change of work function of the films was measured with the Kelvin method, and was compared with the change of resistance when they were exposed to hydrogen. The change of work function in the same concentration of hydrogen shows different trends according to the variation of the film temperature and this result reveals that there exist three different models of hydrogen interaction with the film surface or with the oxygen adsorbates on the surface, which depend on the film temperature at that time. The change of resistance of the films shows quick saturation, while that of the work function has trends of slow and continuous decreasing; from these results, it is concluded that the adsorption or desorption of the oxidizing or reducing gases on the film surface are contributed mainly to the change of potential barrier of grain boundaries of the films.

Indoor Mobile Robot Navigation by Central Following Based on Monocular Vision

This paper develops the indoor mobile robot navigation by center following based on monocular vision. In our method, based on the frontal image, two boundary lines between the wall and baseboard are detected. Then, the appearance based obstacle detection is applied. When the obstacle exists, the avoidance or stop movement is worked according to the size and position of the obstacle, and when the obstacle does not exist, the robot moves at the center of the corridor. We developed the wheelchair based mobile robot. We estimated the accuracy of the boundary line detection, and obtained fast processing speed and high detection accuracy. We demonstrate the effectiveness of our mobile robot by the stopping experiments with various obstacles and moving experiments.

Investigation of Substrate-dependent Characteristics of SnO2 Thin Films with Hall Effect, X-Ray Diffraction, X-Ray Photoelectron Spectroscopy and Atomic Force Microscopy Measurements

SnO2 thin films of 100 nm in thickness were prepared on glass and alumina substrates with targets-facing type sputtering apparatus to investigate the substrate-dependent characteristics of the films. The sensitivity of the films is measured in flammable gas atmosphere (hydrogen, butane and methane, 5000 ppm), and it revealed that the SnO2 films on alumina substrates showed better sensitivity for all introduced gases than the films on glass substrates. Hall effect, X-ray diffraction (XRD), X-ray photoelectron spectroscopy (XPS) and atomic force microscopy (AFM) measurements were performed to clarify the difference of the sensitivity, and it is concluded as follows: 1) The variation of carrier concentration of the films on alumina substrates is larger than those on glass substrates when they are exposed to flammable gases. 2) The structure of the films on alumina substrates is similar to that of SnO2 powder. 3) The film on an alumina substrate contains more oxygen impurities than that on a glass substrate, which can be considered to cause the large variation of carrier concentration. 4) Surface area of the films on alumina substrates is wider than that of the films on glass substrates.

Vowel recognition according to lip shapes by using neural network

Although speech recognition techniques have been successfully developed, the noise of the circumstances causes serious errors in recognition. In such cases, we expect that lip shapes are useful data as supporting data to improve the performance of recognition. We propose a method to extract lip shapes from input face images by using an active contour model. After transformation and normalization the extracted data are input to a neural network to recognize vowels. Experimental results of vowel recognition are shown to confirm effectiveness of the proposed method.

Monocular Autonomy Following Vehicle

This paper describes the development of the monocular autonomy following vehicle which for reduce a helper burden. The system is composed by a CCD camera, a FPGA board and some control circuits. Our idea is that the distance between the system and the target human is related to the size of the human region in the camera image. Based on this idea, we proposed the automatic human region detection method with S-ACM which is a kind of active contour model. We implemented our method and developed the system. The experiment which followed the target human was done. As the result, we confirmed that the system followed with keeping a fixed distance with the target human

IMPROVEMENT OF CONCENTRATION-ESTIMATION ALGORITHM FOR INFLAMMABLE GASES UTILIZING FUZZY RULE-BASED NEURAL NETWORKS

Abstract We have proposed an algorithm which can discriminate inflammable gases and estimate their concentration with a semiconductor gas sensor under the periodic operation in our previous paper. In this paper, we propose fuzzy rule-based neural networks, which are composed of two back propagation neural networks, to improve the estimation accuracy and to reduce the time and efforts for creation and tuning of the membership functions. The proposed network is examined in estimating the concentrations of three kinds of inflammable gases, that is, butane, hydrogen and methane, and it is proved that the results are more accurate than those obtained with simplified fuzzy inference.

Analysis of Gate Lag in GaAs Metal-Semiconductor Field-Effect Transistor Using Light Illumination

We have developed a novel method for analyzing the gate lag effect in gallium arsenide (GaAs) metal-semiconductor field-effect transistor (MESFET) using light illumination. It is estimated that the density of trapped electrons at the surface of an active channel layer is above 6×1011 cm-2 from the dependence on photon flux density. Photon energy dependence shows that the electrons are mainly trapped at the GaAs surface. Angle-resolved analysis indicates that the trapped electrons at the active channel layer between the gate and the drain mostly account for the gate lag effect. Temperature dependence of the transconductance (g m) dispersion shows that the activation energy of this trap is 0.33 eV. Two-dimensional device simulation demonstrates the similar transient characteristics of the drain current, which originates from the electrons trapped at the GaAs surface.