Catálogo de publicaciones - libros

This paper proposes a 3-D position compensation method of robot working points that are used in industrial vehicle assembly lines for an equipment glass. The proposed method uses a laser sensor and a vision camera. Because the laser sensor is a cross type sensor which can scan a horizontal and vertical line, it is an efficient way to detect a feature of vehicle and winding shape of vehicle body. For position compensation of 3-Dimensional axes, we applied block interpolation method using pattern matching method and cross point of laser slit in input image. A 3-D position is selected by Euclidean distance mapping between 315 feature values and evaluated feature points. Because these points are in the robot process program, the processing time is more quickly than that of the computer program. In order to evaluate the proposed algorithm, experiments are performed glass equipment system in real industrial vehicle assembly line. As the result of the field test, it shows that working point of robot can be displayed 3-D points.

A JET-X (Joint European X-ray Telescope) CCD fabricated on a high-purity (1.5 kΩcm), 65 μm thick epi-layer, on a 550 μm thick p substrate has been developed for X-ray astronomy. A three-dimensional (3-D) numerical analysis for evaluating a charge handling capability and charge transfer efficiency of a JET-X CCD using a static and transient simulation has been performed. A supplementary channel technique is analysed by the 3-D simulation. A static maximum charge capacity was found to be 60040 electrons under a full-well condition. The effect of an output gate voltage on charge transfer between the last well and an output diffusion node was observed and an optimum output gate voltage for efficient charge transfer was found to be 3 or 4 V. The 3-D simulated result showed a higher charge capability by appr. 18% than result from a 1-D model.

When a system is teleoperated in the indoor environment through the wireless LAN, the communication time delay that is due to the inherent characteristic and surrounding environment is random and unbounded. The time delay has a significant effect on the stability and performance of the teleoperating system. In this paper, we present the method that is the image compression to reduce the time delay, measuring time delay, and switching control-mode corresponding to time delay automatically, to improve stability and performance. The simple experiment is conducted to demonstrate the feasibility.

In this paper, we proposed a novel pattern classification algorithm based on Minimum Mahalanobis Enclosing Ellipsoid (MMEE) cognitive description for generally imbalanced dataset classification. Following the idea of enclosing machine learning concepts, this method tries to enclose the target class samples while exclude the outlier class samples at the same time. Because this method utilizes the covariance matrix information, and thus the proposed MMEE method is scale-invariant. We also proposed a robust version of MMEE to further enhance its performance. We formulate the MMEE method as a standard Second Order Cone Programming (SOCP) problem. We investigate and compare its performances with standard Support Vector Machines (SVM) and other two hypersphere based kernel methods using real world UCI benchmark datasets for pattern classification.

Aiming at three-dimensional tactile sensing system based on pressure-sensitive conductive rubber, constitutive relation of large deformed incompressible rubber materials and finite element analyzing method are utilized to conduct analog simulation experiment, and finite element model of flexible tactile sensor sensitive unit is established. Nonlinear analyzing technique of ANSYS software is used to analyze sensitive unit structure. The deformation of pressure-sensitive conductive rubber layer is simulated and the feasibility of design scheme is testified through the comparison of simulation data and experiment data.

In this paper, a novel method for independent component analysis (ICA) with 2-D Principle Component Analysis (2DPCA) in face recognition is presented, called 2DPCA-ICA. In this method, 2DPCA is used for dimension reduction, and ICA for recognition. As opposed to the method for ICA based on PCA in face recognition (PCA-ICA), 2DPCA is based on 2-D face image matrix, an image covariance matrix is constructed directly using 2-D image matrix, and its eigenvectors corresponding to the several larger eigenvalues are derived for whitened image matrix. It overcomes the shortcomings of PCA-ICA. To test 2DPCA-ICA and evaluate its performance, experiments are performed on Yale and ORL (Olivetti Research Laboratory) face database. Correct recognition rate of 2DPCA-ICA across all trials is higher than that of PCA-ICA and 2DPCA. Experimental results also show that features of face image extracted are more efficient by way of 2DPCA-ICA. Therefore, 2DPCA-ICA is more valid in face recognition.

Regularization is a key step for solving the ill-posed inverse problem of electrocardiography (ECG). In this paper, a novel regularization technique (LSQR-Tik) which combines the least square QR (LSQR) method with a Tikhonov-like prior information term is proposed. This technique needs to select two parameters, the Tikhonov-like regularization parameter () and the iteration number of LSQR-Tik (), which can be determined by a modified L-curve technique. The performance of the LSQR-Tik method for solving the inverse ECG problem was evaluated based on a realistic heart-torso model simulation protocol. The results show that the LSQR-Tik method could overcome the ill-pose property effectively and get better inverse solutions than those of Tikhonov and LSQR methods, especially in the case of body surface potential with large noises.

Clustering is traditionally viewed as an unsupervised method for data analysis. However, in some cases information about the problem domain is available in addition to the data instances themselves. To make use of this information, in this paper, we develop a new clustering method “MLP-KMEANS” by combining Multi-Layer Perceptron and K-means. We test our method on several data sets with partial constrains available. Experimental results show that our method can effectively improve clustering accuracy by utilizing available information.

In this study, a novel neural network ensemble system, i.e. NNES, is proposed for economic forecasting. This ensemble approach utilizes the boosting and bagging algorithms to constitute neural network ensemble respectively, which are combined into NNES with simple average, and a novel forecasting effective measure algorithm to determine the weights used to form the neural network ensembles. For illustration and testing purposes, the proposed ensemble model is applied for economic forecasting.

This article presents the numerical simulation of the static and dynamic response of viscoelastic materials with the finite element method. In particular, elastomeric solids which are of essential engineering interest are discussed. A rubbery elastic manner is assumed to be modeled with an Ogden-type strain energy function well-known from rubber elasticity. A Prony series represents the relaxation moduli is exploited in order to derive a recursive relationship. By using a general finite element software ANSYS, several representative numerical examples illustrate relaxation and creeping phenomena.