Enzeng Dong

A neural network-based self-tuning PID controller of an autonomous underwater vehicle

Taking into account the complex interferences in underwater environment, this paper presents a neural network-based self-tuning PID controller for a spherical AUV. The control system consists of neural network identifier and neural network controller, and the weights of neural networks are trained by using Davidon least square method. The proposed controller is characterized with a strong anti-interference ability and a fast convergence rate. For its simple structure, the controller can be easily realized in hardware. The linear velocity of the spherical AUV can be controlled to precisely track any desired trajectory in vehicle-fixed coordinate system. The effectiveness of the controller is verified by simulation results.

Realization of biped robot gait planning based on NAO robot development platform

In this paper, the research object is NAO robot, and for the instability problem of motion of biped robot, the NAO robot deep development platform is set up, the prototype and simulator are controlled real-time and the data is collected by using the control platform, So many experiments are completed by the development platform, to improve the stability of the simulator and the prototype. The theoretical value are obtained by the variable length inverted pendulum model, and through the comparison of theoretical and actual value of the joints angle sequence, the validity of method is verified. Firstly, the walking process of robot is analysed, the variable length inverted pendulum model is built; Then combined with the geometric constraints method, according to the NAO robot lower leg intrinsic parameters and experiments parameters, the motion trajectories of centroid and joints angle sequence are obtained respectively by using the kinematics, inverse kinematics and smooth function fitting method; Finally, obtained the joint angle sequence is the premise, in this paper, the stable gait walking is realized respectively on the simulator and NAO robot prototype based on the development platform, compared to the joints angle sequence trajectories of the actual waking and the theory calculation, the validity of the above research is verified.

An EOG signals recognition method based on improved threshold dual tree complex wavelet transform

Human Machine Interface (HMI) system can effectively detect Electrooculogram (EOG) signals of eye movements, extract intension of users, and convert them into control commands of computer or rehabilitation aid devices. Thus, HMI system is easily accepted by the majority of persons with disabilities. Discrete Wavelet Transform (DWT) method was mainly used in feature extraction of EOG signals, but Traditional DWT method was always suffered from severe frequency aliasing and poor shift invariant. In this paper, Dual-tree Complex Wavelet Transform (DTCWT) with a novel threshold calculation method was proposed for feature selection of EOG signals. To verify the proposed method, The EOG signal was collected from 5 normal subjects in the laboratory, and featured selected. Then, Support Vector Machine (SVM) was applied for classification. The average correct detection rate of proposed method was 96.11%, which was higher than Traditional DWT method. These results demonstrate that the DTCWT-SVM algorithm provides high classification accuracy, and suitable for clinical medicine field.

Improved common spatial pattern for brain-computer interfacing

Signal processing of electroencephalography (EEG) plays an important role in brain-computer-interface (BCI) system. It is crucial to select suitable features from EEG signals. This paper proposed a feature selection method which combing independent component analysis (ICA) and common spatial patterns (CSP) to improve the performances of classification. Firstly, EEG signals were filtered with 8-30HZ bandpass filter. Secondly, relative frequency band signals were decomposed into independent components to obtain the solution matrix by ICA, and then the EEG signals were reconstructed from the main components to improve the signal-to-noise ratios. CSP was used to extract the features of EEG signals. Finally, linear discriminate analysis classifier (LDA) and support vector machines (SVM) were used to classify the EEG feature signals. The experiment results showed that the average accuracy achieved by the proposed method were higher, compared common CSP method.

Design and implementation of a moving object tracking system

By combining the classic object detection and tracking algorithms, this paper proposed an automatic detection and tracking algorithm on moving object. The Gaussian mixture model (GMM) is applied to detect object, and the fusion algorithm of Kalman filter and Camshift algorithm is utilized to track object. The Pan/Tile/Zoom (PTZ) control algorithm is used to adjust the PTZ Camera parameters, such as camera rotate and Zoom, which can make the object to locate in the centre of field. The effective of algorithm proposed was verified by hardware experiment platform. The experiment results show that the system designed can automatically detect and track moving object, overcoming the limit of camera view and expanding the scope of tracking to camera. Real-time and accuracy of the system has also been validated.

Classification of four categories of EEG signals based on relevance vector machine

EEG classification is an important signal acquisition equipment of brain-interface research and application. In this paper, an improved relevance vector machine (RVM) is proposed to classify four-class motor imagery EEG signals. The original EEG signals are processed by 3–24Hz band-pass filter. Thereafter, EEG feature vectors are extracted from the band-pass filtered EEG signals with one versus one common spatial patterns (OVO-CSP). Then, the improved RVM algorithm with kernel function, which combines Gaussian kernel function and Cauchy kernel function, is applied to classify the EEG signals. A public dataset (BCI Competition IV-II-a) is employed to verify the proposed improved kernel function method. Five-fold cross-validation was used to ensure that the classification of the experiment is more credible. The classification results of the combined kernel function are compared with results of the Gaussian kernel function and the Cauchy kernel function. The experimental results show that the highest classification accuracy of the proposed kernel function and the single kernel function in the first data set are 64.40% and 60.60%, respectively, and in the second data sets are 67.58% and 63.33%. The average classification accuracy of the proposed kernel function is 4%–4.19% higher than that of the single kernel function. These results show that the proposed method is advantageous in the classification of four-class motor imagery.

An Improved NMS-Based Adaptive Edge Detection Method and Its FPGA Implementation

For improving the processing speed and accuracy of edge detection, an adaptive edge detection method based on improved NMS (nonmaximum suppression) was proposed in this paper. In the method, the gradient image was computed by four directional Sobel operators. Then, the gradient image was processed by using NMS method. By defining a power map function, the elements values of gradient image histogram were mapped into a wider value range. By calculating the maximal between-class variance according to the mapped histogram, the corresponding threshold was obtained as adaptive threshold value in edge detection. Finally, to be convenient for engineering application, the proposed method was realized in FPGA (Field Programmable Gate Array). The experiment results demonstrated that the proposed method was effective in edge detection and suitable for real-time application.

Introducing chaos behavior to kernel relevance vector machine (RVM) for four-class EEG classification

This paper addresses a chaos kernel function for the relevance vector machine (RVM) in EEG signal classification, which is an important component of Brain-Computer Interface (BCI). The novel kernel function has evolved from a chaotic system, which is inspired by the fact that human brain signals depict some chaotic characteristics and behaviors. By introducing the chaotic dynamics to the kernel function, the RVM will be enabled for higher classification capacity. The proposed method is validated within the framework of one versus one common spatial pattern (OVO-CSP) classifier to classify motor imagination (MI) of four movements in a public accessible dataset. To illustrate the performance of the proposed kernel function, Gaussian and Polynomial kernel functions are considered for comparison. Experimental results show that the proposed kernel function achieved higher accuracy than Gaussian and Polynomial kernel functions, which shows that the chaotic behavior consideration is helpful in the EEG signal classification.

Topological Horseshoe Analysis and FPGA Implementation of the Fractional-order Liu System

This paper first discusses a fractional-order Liu system of order as low as 2.7 and shows its chaotic characteristics by carrying out numerical simulations such as Lyapunov exponents, bifurcation diagrams and phase portraits. Then, by using the topological horseshoe theory and computer-assisted proof, the existence of chaos in the system is verified theoretically. Finally, the fractional-order system is implemented on a Field Programmable Gate Array (FPGA) and the results obtained show that the fractional-order Liu system is indeed chaotic.

The multi-frequency EEG rhythms modeling based on two-parameter bifurcation of neural mass model

In this work, a coupled Jansen and Rit neural mass model with two neuron populations were proposed, which was studied further by using the two-parameter bifurcation analysis. The numerical analysis shows that the coupled Jansen and Rit neural mass model can generate a type of multi-frequency rhythms, which are closer proximity to real EEG rhythms. Furthermore, due to the combined effect of two parameters bifurcations, more codimension two bifurcations behaviours on equilibriums and cycles were discovered, which can generate more complex dynamics behaviours. Finally, the validity of the proposed modelling method was verified by the comparative analysis between the simulation EEG rhythms and real EEG rhythms, verified the validity of this system.