Vlastimir Nikolić

Stereo Vision-Based Human Tracking for Robotic Follower

Abstract This paper addresses the problem of real-time vision-based human tracking to enable mobile robots to follow a human co-worker. A novel approach to combine stereo vision-based human detection with human tracking using a modified Kalman filter is presented. Stereo vision-based detection combines features extracted from 2D stereo images with reconstructed 3D object features to detect humans in a robots environment. For human tracking a modified Kalman filter recursively predicts and updates estimates of the 3D coordinates of a human in the robots camera coordinate system. This prediction enables human detection to be performed on the image region of interest contributing to cost effective human tracking. The performance of the presented method was tested within a working scenario of a mobile robot intended to follow a human co-worker in indoor applications as well as in outdoor applications.

Wind speed parameters sensitivity analysis based on fractals and neuro-fuzzy selection technique

Fluctuation of wind speed affects wind energy systems since the potential wind power is proportional the cube of wind speed. Hence precise prediction of wind speed is very important to improve the performances of the systems. Due to unstable behavior of the wind speed above different terrains, in this study fractal characteristics of the wind speed series were analyzed. According to the self-similarity characteristic and the scale invariance, the fractal extrapolate interpolation prediction can be performed by extending the fractal characteristic from internal interval to external interval. Afterward neuro-fuzzy technique was applied to the fractal data because of high nonlinearity of the data. The neuro-fuzzy approach was used to detect the most important variables which affect the wind speed according to the fractal dimensions. The main goal was to investigate the influence of terrain roughness length and different heights of the wind speed on the wind speed prediction.

Potential of support vector regression for optimization of lens system

Lens system design is an important factor in image quality. The main aspect of the lens system design methodology is the optimization procedure. Since optimization is a complex, non-linear task, soft computing optimization algorithms can be used. There are many tools that can be employed to measure optical performance, but the spot diagram is the most useful. The spot diagram gives an indication of the image of a point object. In this paper, the spot size radius is considered an optimization criterion. Intelligent soft computing scheme Support Vector Regression (SVR) is implemented. In this study, the polynomial and radial basis functions (RBF) are applied as the SVR kernel function to estimate the optimal lens system parameters. The performance of the proposed estimators is confirmed with the simulation results. The SVR results are then compared with other soft computing techniques. According to the results, a greater improvement in estimation accuracy can be achieved through the SVR with polynomial basis function compared to other soft computing methodologies. The SVR coefficient of determination R 2 with the polynomial function was 0.9975 and with the radial basis function the R 2 was 0.964. The new optimization methods benefit from the soft computing capabilities of global optimization and multi-objective optimization rather than choosing a starting point by trial and error and combining multiple criteria into a single criterion in conventional lens design techniques. Lens system design represents a crucial factor for good image quality.Optimization procedure is the main part of the lens system design methodology.Soft computing methodologies optimization application.Adaptive neuro-fuzzy inference system (ANFIS) application.Support vector regression (SVR application).

DETERMINATION OF IMPORTANT PARAMETERS FOR PATENT APPLICATIONS

This research study is an analysis of patent applications based on different input parameters. Nine patent indicators for describing patent applications are retrieved from the World Bank database. The method of ANFIS (adaptive neuro fuzzy inference system) is applied to selecting the most important parameters for patent applications. The inputs are: charges for the use of intellectual property for payments and receipts, research and development expenditure, trademark applications for residents and nonresidents, researchers in research and development (R&D), technicians in R&D and high-technology exports. As the ANFIS outputs, patent applications for nonresidents and residents are considered. The results show that the combination of research and development expenditure and technicians in R&D is the most influential combination of input parameters for patent applications.

An approach to neuro-fuzzy filtering for communications and control

To reduce the influence of noise in communication and control systems the use of neuro-fuzzy adaptive nonlinear filters is studied. The most important advantage of the soft computing based adaptive filters is that linguistic and numerical information can be efficiently combined. A unified procedure for filter design is presented, and different approaches are also considered.

Computationally intelligent system for thermal vision people detection and tracking in robotic applications

This paper describes a system for real-time robust segmentation of human in a thermal image used for supervisory control of mobile robot platform. The main goal was to enable mobile robot platform to recognize the person in indoor environment, and to localize it with accuracy high enough to allow adequate human-robot interaction. The developed computationally intelligent control algorithm enables robust and reliable human tracking by mobile robot platform. The core of the recognition methods proposed is intelligent segmentation and classification of detected regions of interests in every frame acquired by thermal vision camera. Advanced intelligent segmentation algorithm is based on improved fuzzy closed-loop colour region segmentation. This segmentation algorithm enables autonomous functioning of robot system in cluttered environments. The classifier determines whether the segmented object is human or not based on features extracted from the processed thermal image. With this approach a person can be detected independently from current light conditions and in situations where no skin colour is visible. However, variation in temperature across same objects, air flow with different temperature gradients, person overlap while crossing each other and reflections, put challenges in thermal imaging and will have to be handled intelligently in order to obtain the efficient performance from motion tracking system. Presented research in this field includes making tracking system more robust and reliable by using the computational intelligence.

A NEW APPROACH TO THE SLIDING MODE CONTROL DESIGN: ANTI–LOCK BRAKING SYSTEM AS A CASE STUDY

Abstract In this paper we introduce a new approach to the sliding mode control design based on orthogonal models. First, we discuss the sliding mode control based on a model given in controllable canonical form. Then, we design almost orthogonal filters based on almost orthogonal polynomials of M¨untz-Legendre type. The advantage of the almost orthogonal filters is that they can be used for the modelling and analysis of systems with nonlinearities and imperfections. Herein, we use a designed filter to obtain several linearized models of an unknown system in different working areas. For each of these linearized models, corresponding sliding mode controller is designed and the switching between controls laws depends only on input signal. The experimental results and comparative analysis with relay control, already installed in laboratory equipment, verify the efficiency and excellent performance of such a control in the case of anti-lock braking system.

Near field acoustic localization under unfavorable conditions using feedforward neural network for processing time difference of arrival

Providing guidelines for practical implementation of neural networks in near-field sound source localization.Obtained optimal sensors setups.Obtaining optimal network configuration.Obtaining optimal training parameters.Proving effectiveness of feedforward neural network in solving hyperbolic positioning problem under the uncertainties. Using time difference of arrival (TDOA) is one of the two approaches that utilize time delay for acoustic source localization. Combining the obtained TDOAs together with geometrical relationships within acoustic components results in a system of hyperbolic equations. Solving these hyperbolic equations is not a trivial procedure especially in the case of a large number of microphones. The solution is additionally compounded by uncertainties of different backgrounds. The paper investigates the performance of neural networks in modelling a hyperbolic positioning problem using a feedforward neural network as a representative. For experimental purposes, more than 2000 sound files were recorded by 8 spatially disposed microphones, for as many arbitrarily chosen acoustic source positions. The samples were corrupted by high level correlated noise and reverberation. Using cross-correlation, with previous signal pre-processing, TDOAs were evaluated for every pair of microphones. On the basis of the obtained TDOAs and accurate sound source positions, the neural network was trained to perform sound source localization. The performance was examined using a large number of samples in terms of different acoustic sensors setups, network configurations and training parameters. The experiment provided useful guidelines for the practical implementation of feedforward neural networks in the near-field acoustic localization. The procedure does not require substantial knowledge of signal processing and that is why it is suitable for a broad range of users.

Vibration prediction of pellet mills power transmission by artificial neural network

Purpose The purpose of this study is to establish a vibration prediction of pellet mills power transmission by artificial neural network. Vibration monitoring is an important task for any system to ensure safe operations. Improvement of control strategies is crucial for the vibration monitoring. Design/methodology/approach As predictive control is one of the options for the vibration monitoring in this paper, the predictive model for vibration monitoring was created. Findings Although the achieved prediction results were acceptable, there is need for more work to apply and test these results in real environment. Originality/value Artificial neural network (ANN) was implemented as the predictive model while extreme learning machine (ELM) and back propagation (BP) learning schemes were used as training algorithms for the ANN. BP learning algorithm minimizes the error function by using the gradient descent method. ELM training algorithm is based on selecting of the input weights randomly of the ANN network and the output weight of the network are determined analytically.

FUZZY CONTROL OF DIFFERENTIAL DRIVE MOBILE ROBOT FOR MOVING TARGET TRACKING

Tracking of moving objects, including humans has important role in mobile robotics. In this paper, the hierarchical control structure for target/human tracking consisted of high and low level control was presented. The low level subsystem deals with the control of the linear and angular velocities using multivariable PD controller whose parameters are obtained by Particle swarm optimization. The position control of the mobile robot represents the high level control, where we use two fuzzy logic Mamdani controllers for distance and angle control. In order to test the effectiveness of the proposed control scheme a simulation was performed. Two cases, when the mobile robot pursues a target moving along a circular path and when the mobile robot pursues a target moving along a rectangle path, were simulated.