Showing papers by "Zeng-Guang Hou published in 2008"

Decentralized adaptive leader-follower control of multi-manipulator system with uncertain dynamics

Abstract: An adaptive control approach is proposed to deal with the multi-manipulator leader-follower control based on the multi-agent theory In the current multi-agent literature, agents are assumed to have determined linear models However, the real manipulatorpsilas dynamics is highly nonlinear and contains uncertain parameters According to the ldquoinearity-in-parametersrdquo property, the adaptive updating law for uncertain dynamics parameters is derived Then, the decentralized torque controller for each follower manipulator is designed using the backstepping scheme, which only utilizes the information of its neighbor (connected) manipulators By the proposed controller, all the follower manipulatorspsila joints can track the leader jointpsilas trajectory to achieve certain coordination tasks In addition, the performance of control system is analyzed by the Lyapunov method, and the tracking error is proved to approach zero Finally, the effectiveness of the proposed scheme is illustrated by simulations on a multiple two-link manipulators system

Fuzzy Control of the Inspection Robot for Obstacle-Negotiation

Abstract: This paper proposes the structure of hierarchical behavior controller for the inspection robot. And a fuzzy controller is designed based on the structure to realize the turning behavior of robot. The inspection robot is a multivariable, nonlinear, strongly coupling and underactuated dynamic system. It is difficult to establish an accurate mathematic model of the system. The fuzzy control method can be applicable to those objects which hardly have an accurate mathematic model or whose system parameters are variable. This paper also discusses the stability of fuzzy control system. The experiment results show that the designed fuzzy controller can meet the requirement of the obstacle-negotiation control system.

Adaptive neural network tracking control of manipulators using quaternion feedback

Abstract: An adaptive neural network controller is proposed to deal with the task-space tracking problem of manipulators with kinematic and dynamic uncertainties. The orientation of manipulator is represented by the unit quaternion, which avoids singularities associated with three-parameter representation. By employing the adaptive Jacobian scheme, neural networks, and backstepping technique, the torque controller is obtained which is demonstrated to be stable by the Lyapunov approach. The adaptive updating laws for controller parameters are derived by the projection method, and the tracking error can be reduced as small as desired. The favorable features of the proposed controller lie in that: (1) the uncertainty in manipulator kinematics is taken into account; (2) the unit quaternion is used to represent the end-effector orientation; (3) the "linearity-in-parameters" assumption for the uncertain terms in dynamics of manipulators is no longer necessary; (4) effects of external disturbances are also considered in the controller design. Finally, the satisfactory performance of the proposed approach is illustrated by simulation results on a PUMA 560 robot.

A systematic topological method for fingerprint singular point detection

Abstract: Singular point detection is an important issue in fingerprint image analysis. General methods like Poincare index method can detect singular points in non-arch type fingerprints but fail on arch-type fingerprints. Some more sophisticated methods like complex filter method also face the same problem. In this paper, we propose a systematic method for detecting singular points in fingerprint images which utilizes the most fundamental topological feature of acquired fingerprints as the basis for singular point identification. The method differentiates the input fingerprint between arch type and non-arch type. For non-arch type fingerprints singular points are detected as intersection points in a c(>2) level segmentation map. As for arch type fingerprints, singular points are identified from the symmetric line of the fingerprint structure. The method is evaluated using the NIST DB4 database and compared with the complex filter method. The proposed method attains near 90% success rate in detecting singular points and the displacement from ground truth is comparable to that of the complex filter method. Moreover our method is more than once faster in CPU time.

Online mapping with a mobile robot in dynamic and unknown environments

Abstract: In this paper, we address the problem of mapping dynamic and unknown environments The static and moving objects are modelled as the components in a Gaussian mixture model (GMM) By recursive learning of GMM, the components corresponding to the static objects will have larger weights while the components corresponding to the moving objects will have smaller weights At each time step, a number of components with the largest weights are adaptively selected as the background map and the new observations which do not match with the background map are classified as the foreground map In addition, based on a Bayesian factorisation of simultaneous localisation and mapping (SLAM) problem, we present an online algorithm for SLAM with GMM learning Our contributions are employing GMM learning approach to model the dynamic environment with detection of moving objects and jointing the GMM learning with SLAM in unknown environment Consequently, an online approach for mapping with a mobile robot in dynamic and unknown environments is presented Some simulation results indicate that our approach is feasible

Multiple kernel learning from sets of partially matching image features

Abstract: Recent publications and developments based on SVM have shown that using multiple kernels instead of a single one can enhance interpretability of the decision function and improve classifier performance, which motivates researchers to explore the use of homogeneous model obtained as linear combinations of kernels. However, the use of multiple kernels faces the challenge of choosing the kernel weights, and an increased number of parameters that may lead to overfitting. In this paper we show that MKL problem with a enhanced spatial pyramid match kernel can be solved efficiently using projected gradient method. Weights on each kernel matrix (level) are included in the standard SVM empirical risk minimization problem with a L2 constraint to encourage sparsity. We demonstrate our algorithm on classification tasks, which is based on a linear combination of the proposed kernels computed at multiple pyramid levels of image encoding, and we show that the proposed method is accurate and significantly more efficient than current approaches.

Corridor-Scene Classification for Mobile Robot Using Spiking Neurons

Abstract: The ability of cognition and recognition for complex environment is very important for a real autonomous robot. A corridor-scene-classifier based on spiking neural networks (SNN) for mobile robot is designed to help the mobile robot to locate correctly. In the SNN classifier, the integrate-and-fire model (IAF) spiking neuron model is used and there is lateral inhibiting in the output layer. The winner-take-all rule is used to modify the connecting weights between the hidden layer and the outputting layer. The experimental results show that the corridor-scene-classifier is effective and it also has strong robustness.

Vision based navigation for power transmission line inspection robot

Abstract: Inspection robot must plan its behavior to loose or grasp the power transmission line, or recognize the obstacles from the complex background when it is crawling along the line in order to negotiate reliably. This paper describes a vision-based navigation system for a power line inspection robot. The main emphasis of this paper is on the ability of object recognition. A recognition method based on straight line extraction is proposed, which is used to recognize the typical obstacles in the power transmission line. Random sample consensus (RANSAC) paradigm is used to group the line segments. The proposed method scales well with respect to the size of the input image and the number and size of the shapes within the data. Moreover the algorithm is conceptually simple and easy to implement. Experimental results show that good recognition can be achieved using the proposed vision system.

Unsupervised learning of categories from sets of partially matching image features for power line inspection robot

Abstract: Object recognition and categorization are considered as fundamental steps in the vision based navigation for inspection robot as it must plan its behaviors based on various kinds of obstacles detected from the complex background. However, current approaches typically require some amount of supervision, which is viewed as a expensive burden and restricted to relatively small number of applications in practice. For this purpose, we present an computationally efficient approach that does not need supervision and is capable of learning object categories automatically from unlabeled images which are represented by an set of local features, and all sets are clustered according to their partial-match feature correspondences, which is done by a enhanced Spatial Pyramid Match algorithm (E-SPK). Then a graph-theoretic clustering method is applied to seek the primary grouping among the images. The consistent subsets within the groups are identified by inferring category templates. Given the input, the output of the approach is a partition of the images into a set of learned categories. We demonstrate this approach on a field experiment for a powerline inspection robot.

Foundations of Nonconventional Neural Units and their Classification

Abstract: This article introduces basic types of nonconventional neural units and focuses on their notation and classification. Namely, the notation and classification of higher order nonlinear neural units, time-delay dynamic neural units, and time-delay higher order nonlinear neural units are introduced. Brief introduction into the simplified parallels of the higher order nonlinear aggregating function of higher order neural units with both the synaptic and somatic neural operation of biological neurons is made. Based on the mathematical notation of neural input intercorrelations of higher order neural units, it is shown that the higher order polynomial aggregating function of neural inputs can be understood as a single-equation representation of synaptic neural operation plus partial somatic neural operation. Thus, it unravels new simplified yet universal mathematical insight into understanding the higher computational power of neurons that also conforms to biological neuronal morphology. The classification of nonconventional neural units is founded first according to the nonlinearity of the aggregating function; second, according to the dynamic order; and third, according to time-delay implementation within neural units.

A Multi-agent Architecture Based Cooperation and Intelligent Decision Making Method for Multirobot Systems

Abstract: The design of a hybrid multi-agent architecture is proposed for multirobot systems. Analysis of the architecture shows that it is suitable for multirobot systems dealing with changing environments. Meanwhile, it is capable of controlling a group of robots to accomplish multiple tasks simultaneously. Two associated issues about the architecture are cooperation between robots and intelligent decision making. Ability vector, cost function and reward function are used as criteria to describe and solve the role assignment problem in multirobot cooperation. A solution of information fusion based on RBF neural networks is applied to solve the reality problem in decision making of multirobot systems. And an experiment about robot soccer shooting is designed. The experimental results verify that the method can improve the whole decision system in accuracy.

Adaptive dynamic surface control of a class of uncertain nonlinear systems in pure-feedback form using fuzzy backstepping approach

Abstract: This paper presents an adaptive fuzzy controller for the robust backstepping control of a class of uncertain nonlinear systems in pure-feedback form. The proposed approach incorporates the Nussbaum gain function (NGF) and the dynamic surface control (DSC) into the existing adaptive fuzzy control scheme. The major features of the proposed method are that: 1) the two problems, control directions and control singularity, are well solved by use of the NGF; 2) the problem of ldquoexplosion of complexityrdquo inherent to the existing backstepping methods is eliminated by introducing the DSC technique; 3) the proposed control scheme has the adaptive mechanism with fewer adaptive laws, which results in less computation time burden. The performance of the proposed approach is demonstrated through a simulation example.

Nonlinear Calibration of pH Sensor Based on the Back-propagation Neural Network

Abstract: The relationship between pH and voltage output of the pH sensor with glass electrode is nonlinear. To improve the accuracy of the pH sensor, many algorithms have been proposed. This paper mainly focuses on nonlinear calibration of pH sensor using Back-propagation neural network. The operating principle of the glass electrode is introduced in detail, and then, the methods and the principles of sensor calibration are discussed. The experimental results demonstrate that the method is efficient and the system is reliable.

Neurodynamic programming: a case study of the traveling salesman problem

Abstract: The paper focuses on the study of solving the large-scale traveling salesman problem (TSP) based on neurodynamic programming. From this perspective, two methods, temporal difference learning and approximate Sarsa, are presented in detail. In essence, both of them try to learn an appropriate evaluation function on the basis of a finite amount of experience. To evaluate their performances, some computational experiments on both the Euclidean and asymmetric TSP instances are conducted. In contrast with the large size of the state space, only a few training sets have been used to obtain the initial results. Hence, the results are acceptable and encouraging in comparisons with some classical algorithms, and further study of this kind of methods, as well as applications in combinatorial optimization problems, is worth investigating.

A simplified recurrent neural network for solving nonlinear variational inequalities

Abstract: A recurrent neural network is proposed to deal with the nonlinear variational inequalities with linear equality and nonlinear inequality constraints. By exploiting the equality constraints, the original variational inequality problem can be transformed into a simplified one with only inequality constraints. Therefore, by solving this simplified problem, the neural network architecture complexity is reduced dramatically. In addition, the proposed neural network can also be applied to the constrained optimization problems, and it is proved that the convex condition on the objective function of the optimization problem can be relaxed. Finally, the satisfactory performance of the proposed approach is demonstrated by simulation examples.

Forward Passageway based collision-free target tracking for mobile robot with local sensing

Abstract: This paper proposes a new forward passageway (FP) based real-time collision-free target tracking approach for a mobile robot with local sensing. After the position of the target is estimated and localized in robot coordinate system through the combination of vision system and encoder, the sonar information and the target position are converted to a uniform environment model framework called decision-making space. Based on the space, a FP based decision-making is given to endow the robot with the ability to avoid possible obstacles and track the target in unknown environments. Experiment results show the validity of the proposed approach.

Dual Heuristic Programming Based Neurocontroller for Vibration Isolation Control

Abstract: The dual heuristic programming (DHP) approach has a superior capability to solve approximate dynamic programming problems in the family of adaptive critic designs (ACDs). In this paper, a DHP based controller is developed for vibration isolation applications. In the specific case of an active-passive isolator, a multilayer feedforward neural network is pre- trained as a differentiable model of the plant for the adaptation of the critic and action networks. In addition, in order to avoid plunging in the local minima during the training process, pseudorandom signals, which represent the vibrations of the base, are applied to the vibration isolation system. This technique greatly improves the robustness of the DHP controller against unknown disturbances. Moreover, the "shadow critic" training strategy is adopted to improve the convergence rate of the training. Simulation results have shown that the developed DHP controller alleviates vibration disturbances more effectively and have better control performance in comparison with the passive isolator. Additionally, as compared with the one adapted on-line, the pre-trained model network leads the training to be more efficient. Therefore, it further demonstrates the effectiveness of the DHP controller designed with the pre-trained model network even in the presence of unmodeled uncertainties.

Fingerprint orientation analysis with topological modeling

Abstract: This paper proposes to characterize the fingerprint orientation using a novel topological representation, which transforms the orientation field into a map composed of a set of geometric objects and analyzes the properties of each geometric object based on two assumptions: regional coherence assumption and convexity assumption. Different from prior works on fingerprint orientation analysis, this approach has a capability to identify where the errors happen, not only the ldquorandom errorsrdquo (which is assumed by local-filtering and global modeling methods), but also the ldquostructure irregularitiesrdquo, which often occur in real prints and impose difficulty on existing techniques.

Editorial to Special Issue: Neural networks for pattern recognition and data mining

Abstract: The Fourth International Symposium onNeural Networks (ISNN’07) was held in Nanjing, China, between June 3 and June 7, 2007. The ISNN’07 was a great success and covered broad topics of neural network research and applications in diverse fields. As a continuation of this successful conference,weorganized this SoftComputing journal special issue on “neural networks for pattern recognition anddatamining”, with 13 selected high-quality papers from ISNN’07. The theme of this special issue is to present the stateof-the-art developments in recent research focusing on neural networks for pattern recognition and data mining. Over the past decades, many important theoretical and application researches were developed in the area of neural networks and learning systems. Although the understanding of natural intelligent behavior and design brain-like intelligent machines still remains a challengic topic, neural networks have proven successful in many real-world applications. To this end, the selected 13 papers in this special issue can be categorized into four major sections under this theme. The first section is focused on the neural network models and applications. For instance, V. Bevilacqua et al. presented the research on face recognition based on pseudo-2DHidden Markov Models (HMMs). In this approach, each HMM is trained by coefficients of a neural network. Simulation results on the Olivetti Research Laboratory face database illustrated

Application of an adaptive extended Kalman filter in SINS/GPS integrated navigation system

Abstract: For low-cost SINS/GPS integrated navigation system using low precision inertial sensors, the psi-angle model of SINS will degrade the performance of a designed filter due to the neglected error terms in the model. A general SINS error propagation model based on quaternion is presented. The error model is nonlinear and does not rely on small misalignment angles assumption. To estimate SINS errors with uncertain measurement noise, an adaptive extended Kalman filtering method is proposed. In the adaptive filtering method, the components of a measurement vector are processed serially, and the measurement noise covariance matrix is estimated on-line. The simulation results reveal that without exact process and measurement noise covariance matrixes, the adaptive extended Kalman filtering approach can improve the performance of low-cost SINS/GPS integrated navigation system effectively.

Heuristic dynamic programming for semi-active control of vibration isolation

Abstract: Vibration isolation controllers are used to suppress undesirable disturbance in order to guarantee better performance in many industrial and scientific domains. To overcome the drawbacks of the conventional passive systems, a novel design based on the action dependent heuristic dynamic programming (ADHDP) is addressed in this paper for the semi-active vibration isolator. ADHDP, derived from dynamic programming, is adopted in the design of a nonlinear optimal vibration controller. This approach is the simplest category of adaptive critic designs (ACDs) which are very efficient to solve a class of nonlinear optimal control problems. Only two subnetworks are involved in ADHDP, namely the action network and the critic network. Least mean square (LMS) algorithm with variable learning rate is applied to the adaptation of these two networks. A single-stage training process is also demonstrated as a useful training strategy. Two types of vibrations are utilized to verify the effectiveness of this control design. Simulation results show that the semi-active vibration isolator is able to reduce the influence of the vibration excitation to the payload significantly in comparison with the passive system.

Signal Recovery Using Neural Networks for DSP Based Infrared Gas Sensor

Abstract: This paper mainly focuses on signal recovery using neural networks (NN) for digital signal processor (DSP) based infrared gas sensor. Data processing of the infrared gas sensor is discussed in detail, and then, the methods of solving two problems of the data processing are introduced. Finally, experiment results are presented to show the feasibilities and effectiveness of recovering signals using NN. The advantages and challenge of using NN in the infrared gas sensor are also analyzed.

Shape features extraction from pulmonary nodules in X-ray CT images

Abstract: In this paper, we propose a new computer aided diagnosis method of pulmonary nodules in X-ray CT images to reduce false positive (FP) rate under high true positive (TP) rate conditions. An essential core of the method is to extract and combine two novel and effective features from the raw CT images: One is orientation features of nodules in a region of interest (ROI) extracted by a Gabor filter, while the other is variation of CT values of the ROI in the direction along body axis. By using the extracted features, a principal component analysis technic and any pattern recognition technics such as neural network approaches can then used to discriminate between nodule and non-nodule images. Simulation results show that discrimination performance using the proposed features is extremely improved compared to that of the conventional method.

Local sensing based hunting for multiple robots

Abstract: This paper proposes a hunting control approach for multiple mobile robots with local sensing. Predator Robot (PR) requires the sensing information and makes decision without communication with other PRs. The cooperation may emerge by local interactions among the robots. The invader (IR) is given the intelligent ability to escape. Experiments results show the validity of the proposed approach.