In generative dialog systems, learning representations for the dialog context is a crucial step in generating high quality responses. The dialog systems are required to capture useful and compact information from mutually dependent sentences such that the generation process can effectively attend to the central semantics. Unfortunately, existing methods may not effectively identify importance distributions for each lower position when computing an upper level feature, which may lead to the loss of information critical to the constitution of the final context representations. To address this issue, we propose a transfer learning based method named transfer hierarchical attention network (THAN). The THAN model can leverage useful prior knowledge from two related auxiliary tasks, i.e., keyword extraction and sentence entailment, to facilitate the dialog representation learning for the main dialog generation task. During the transfer process, the syntactic structure and semantic relationship from the auxiliary tasks are distilled to enhance both the word-level and sentence-level attention mechanisms for the dialog system. Empirically, extensive experiments on the Twitter Dialog Corpus and the PERSONA-CHAT dataset demonstrate the effectiveness of the proposed THAN model compared with the state-of-the-art methods.
Three-dimensional (3D) visual tracking of a multicopter (where the camera is fixed while the multicopter is moving) means continuously recovering the six-degree-of-freedom pose of the multicopter relative to the camera. It can be used in many applications, such as precision terminal guidance and control algorithm validation for multicopters. However, it is difficult for many researchers to build a 3D visual tracking system for multicopters (VTSMs) by using cheap and off-the-shelf cameras. This paper firstly gives an overview of the three key technologies of a 3D VTSMs: multi-camera placement, multi-camera calibration and pose estimation for multicopters. Then, some representative 3D visual tracking systems for multicopters are introduced. Finally, the future development of the 3D VTSMs is analyzed and summarized.
Aiming to differentiate between mild cognitive impairment (MCI) patients and elderly control subjects, this study proposes an integrated framework based on spectral-temporal analysis for the automatic analysis of resting-state electroencephalogram (EEG) recordings. This framework firstly eliminates noise by employing stationary wavelet transformation (SWT). Then, a set of features is extracted through spectral-temporal analysis. Next, a new wrapper algorithm, named three-dimensional (3-D) evaluation algorithm, is proposed to derive an optimal feature subset. Finally, the support vector machine (SVM) algorithm is adopted to identify MCI patients on the optimal feature subset. Decision tree and K-nearest neighbors (KNN) algorithms are also used to test the effectiveness of the selected feature subset. Twenty-two subjects are involved in experiments, of which eleven persons were in an MCI condition and the rest were elderly control subjects. Extensive experiments show that our method is able to classify MCI patients and elderly control subjects automatically and effectively, with the accuracy of 96.94% achieved by the SVM classifier. Decision tree and KNN algorithms also achieved superior results based on the optimal feature subset extracted by the proposed framework. This study is conducive to timely diagnosis and intervention for MCI patients, and therefore to delaying cognitive decline and dementia onset.
Pneumatic artificial muscles (PAM) have been recently considered as a prominent challenge regarding pneumatic actuators specifically for rehabilitation and medical applications. Since accomplishing accurate control of the PAM is comparatively complicated due to time-varying behavior, elasticity and ambiguous characteristics, a high performance and efficient control approach should be adopted. Besides of the mentioned challenges, limited course length is another predicament with the PAM control. In this regard, this paper proposes a new hybrid dynamic neural network (DNN) and proportional integral derivative (PID) controller for the position of the PAM. In order to enhance the proficiency of the controller, the problem under study is designed in the form of an optimization trend. Considering the potential of particle swarm optimization, it has been applied to optimally tune the PID-DNN parameters. To verify the performance of the proposed controller, it has been implemented on a real-time system and compared to a conventional sliding mode controller. Simulation and experimental results show the effectiveness of the proposed controller in tracking the reference signals in the entire course of the PAM.
The return capsule needs to be launched to the moon and return back to earth in the third stage of the Chinese lunar exploration project. Therefore, it is necessary to perform simulations on the ground. This paper presents an 8-cable-driven parallel manipulator to achieve end-force output in a low-gravity environment. End-force output refers to the vector sum of the external force on the end-effector. A model of end-force output is established based on a kinematics model, a dynamic model, and a force analysis of an 8-cable driven parallel manipulator. To obtain end-force output in a low-gravity environment, the cable force has to be controlled to counteract gravity. In addition, a force-position mix control strategy is proposed to proactively control the cable force according to the force optimal distribution given by the closed-form force distribution method. Furthermore, a suitable choice for an end-force output is obtained by modeling the effect of cable force on end-force output. Experimental results show that the actual cable force agrees well with the calculated force distribution, indicating that it is feasible to realize end-force output in a low gravity environment.
Dynamic hand gesture recognition is a desired alternative means for human-computer interactions. This paper presents a hand gesture recognition system that is designed for the control of flights of unmanned aerial vehicles (UAV). A data representation model that represents a dynamic gesture sequence by converting the 4-D spatiotemporal data to 2-D matrix and a 1-D array is introduced. To train the system to recognize designed gestures, skeleton data collected from a Leap Motion Controller are converted to two different data models. As many as 9 124 samples of the training dataset, 1 938 samples of the testing dataset are created to train and test the proposed three deep learning neural networks, which are a 2-layer fully connected neural network, a 5-layer fully connected neural network and an 8-layer convolutional neural network. The static testing results show that the 2-layer fully connected neural network achieves an average accuracy of 96.7% on scaled datasets and 12.3% on non-scaled datasets. The 5-layer fully connected neural network achieves an average accuracy of 98.0% on scaled datasets and 89.1% on non-scaled datasets. The 8-layer convolutional neural network achieves an average accuracy of 89.6% on scaled datasets and 96.9% on non-scaled datasets. Testing on a drone-kit simulator and a real drone shows that this system is feasible for drone flight controls.
Hyperspectral images in remote sensing include hundreds of spectral bands that provide valuable information for accurately identify objects. In this paper, a new method of classifying hyperspectral images using spectral spatial information has been presented. Here, using the hyperspectral signal subspace identification (HYSIME) method which estimates the signal and noise correlation matrix and selects a subset of eigenvalues for the best representation of the signal subspace in order to minimize the mean square error, subsets from the main sample space have been extracted. After subspace extraction with the help of the HYSIME method, the edge-preserving filtering (EPF), and classification of the hyperspectral subspace using a support vector machine (SVM), results were then merged into the decision-making level using majority rule to create the spectral-spatial classifier. The simulation results showed that the spectral-spatial classifier presented leads to significant improvement in the accuracy and validity of the classification of Indiana, Pavia and Salinas hyperspectral images, such that it can classify these images with 98.79%, 98.88% and 97.31% accuracy, respectively.
Tracking registration is a key issue in augmented reality applications, particularly where there are no artificial identifier placed manually. In this paper, an efficient markerless tracking registration algorithm which combines the detector and the tracker is presented for the augmented reality system. We capture the target images in real scenes as template images, use the random ferns classifier for target detection and solve the problem of reinitialization after tracking registration failures due to changes in ambient lighting or occlusion of targets. Once the target has been successfully detected, the pyramid Lucas-Kanade (LK) optical flow tracker is used to track the detected target in real time to solve the problem of slow speed. The least median of squares (LMedS) method is used to adaptively calculate the homography matrix, and then the three-dimensional pose is estimated and the virtual object is rendered and registered. Experimental results demonstrate that the algorithm is more accurate, faster and more robust.
Based on Lie group theory and the integration of configuration, a class of 3T2R
This paper investigates the stabilisation problem and consider transient optimisation for a class of the multi-input-multi-output (MIMO) semi-linear stochastic systems. A control algorithm is presented via an m-block backstepping controller design where the closed-loop system has been stabilized in a probabilistic sense and the transient performance is optimisable by optimised by searching the design parameters under the given criterion. In particular, the transient randomness and the probabilistic decoupling will be investigated as case studies. Note that the presented control algorithm can be potentially extended as a framework based on the various performance criteria. To evaluate the effectiveness of this proposed control framework, a numerical example is given with simulation results. In summary, the key contributions of this paper are stated as follows: 1) one block backstepping-based output feedback control design is developed to stabilize the dynamic MIMO semi-linear stochastic systems using a linear estimator; 2) the randomness and probabilistic couplings of the system outputs have been minimized based on the optimisation of the design parameters of the controller; 3) a control framework with transient performance enhancement of multi-variable semi-linear stochastic systems has been discussed.
This article proposes a simultaneous localization and mapping (SLAM) version with continuous probabilistic mapping (CP-SLAM), i.e., an algorithm of simultaneous localization and mapping that avoids the use of grids, and thus, does not require a discretized environment. A Markov random field (MRF) is considered to model this SLAM version with high spatial resolution maps. The mapping methodology is based on a point cloud generated by successive observations of the environment, which is kept bounded and representative by including a novel recursive subsampling method. The CP-SLAM problem is solved via iterated conditional modes (ICM), which is a classic algorithm with theoretical convergence over any MRF. The probabilistic maps are the most appropriate to represent dynamic environments, and can be easily implemented in other versions of the SLAM problem, such as the multi-robot version. Simulations and real experiments show the flexibility and excellent performance of this proposal.
Multi-objective dimensional optimization of parallel kinematic manipulators (PKMs) remains a challenging and worthwhile research endeavor. This paper presents a straightforward and systematic methodology for implementing the structure optimization analysis of a 3-prismatic-universal-universal (PUU) PKM when simultaneously considering motion transmission, velocity transmission and acceleration transmission. Firstly, inspired by a planar four-bar linkage mechanism, the motion transmission index of the spatial parallel manipulator is based on transmission angle which is defined as the pressure angle amongst limbs. Then, the velocity transmission index and acceleration transmission index are derived through the corresponding kinematics model. The multi-objective dimensional optimization under specific constraints is carried out by the improved non-dominated sorting genetic algorithm (NSGA II), resulting in a set of Pareto optimal solutions. The final chosen solution shows that the manipulator with the optimized structure parameters can provide excellent motion, velocity and acceleration transmission properties.
The research deals with the methodology intended to root robust quality indices in the interval control system, the parameters of which are affinely included in the coefficients of a characteristic polynomial. To determine the root quality indices we propose to depict on the root plane not all edges of the interval parametric polytope (as the edge theorem says), but its particular vertex-edge route. In order to define this route we need to know the angle sequence at which the edge branches depart from any integrated pole on the allocation area. It is revealed that the edge branches can integrate into the route both fully or partially due to intersection with other branches. The conditions which determine the intersection of one-face edge images have been proven. It is shown that the root quality indices can be determined by its ends or by any other internal point depending on a type of edge branch. The conditions which allow determining the edge branch type have been identified. On the basis of these studies we developed the algorithm intended to construct a boundary vertex-edge route on the polytope with the interval parameters of the system. As an illustration of how the algorithm can be implemented, we determined and introduced the root indices reflecting the robust quality of the system used to stabilize the position of an underwater charging station for autonomous unmanned vehicles.
The multi-robot systems (MRS) exploration and fire searching problem is an important application of mobile robots which require massive computation capability that exceeds the ability of traditional MRS′s. This paper propose a cloud-based hybrid decentralized partially observable semi-Markov decision process (HDec-POSMDPs) model. The proposed model is implemented for MRS exploration and fire searching application based on the Internet of things (IoT) cloud robotics framework. In this implementation the heavy and expensive computational tasks are offloaded to the cloud servers. The proposed model achieves a significant improvement in the computation burden of the whole task relative to a traditional MRS. The proposed model is applied to explore and search for fire objects in an unknown environment; using different sets of robots sizes. The preliminary evaluation of this implementation demonstrates that as the parallelism of computational instances increase the delay of new actuation commands which will be decreased, the mean time of task completion is decreased, the number of turns in the path from the start pose cells to the target cells is minimized and the energy consumption for each robot is reduced.
In this paper, the problem of load transportation and robust mitigation of payload oscillations in uncertain tower-cranes is addressed. This problem is tackled through a control scheme based on the philosophy of active-disturbance-rejection. Here, a general disturbance model built with two dominant components: polynomial and harmonic, is stated. Then, a disturbance observer is formulated through state-vector augmentation of the tower-crane model. Thus, better performance of estimations for system states and disturbances is achieved. The control law is then formulated to actively reject the disturbances but also to accommodate the closed-loop system dynamics even under system uncertainty. The proposed control schema is validated via experimentation using a small-scale tower-crane, and compared with other relevant active disturbance rejection control (ADRC)-based techniques. The experimental results show that the proposed control scheme is robust under parametric uncertainty of the system, and provides improved attenuation of payload oscillations even under system uncertainty.
In this paper, we propose a method to select the observation position in visual servoing with an eye-in-vehicle configuration for the manipulator. In traditional visual servoing, the images taken by the camera may have various problems, including being out of view, large perspective aberrance, improper projection area of object in images and so on. In this paper, we propose a method to determine the observation position to solve these problems. A mobile robot system with pan-tilt camera is designed, which calculates the observation position based on an observation and then moves there. Both simulation and experimental results are provided to validate the effectiveness of the proposed method.
This paper addresses an advanced analysis system for the identification of alcoholic brain states from electroencephalogram (EEG) data in an automatic way. This study introduces an optimum allocation based sampling (OAS) scheme to discover the most favourable representative data points from every single time-window of each EEG signal considering the minimal variability of the observations. Combining all representative samples of each time-window in a set, some statistical features are extracted from every set of each class. The Mann-Whitney U test is used to assess whether each of the features is significant between the two classes (e.g., alcoholic and control). In order to evaluate the effectiveness of the OAS-based features, four well-known machine learning methods (decision table, support vector machine (SVM), k-nearest neighbor (k-NN) and logistic regression) are considered for identification of alcoholic brain state. The experimental results on the UCI KDD (i.e., UCI knowledge discovery in databases) database demonstrate that the OAS based decision table algorithm yields the highest accuracy of 99.58% with a low false alarm rate 0.40%, which is an improvement of up to 9.58% over the existing algorithms. A proposed analysis system can be used to detect alcoholism and also to determine the level of alcoholism-related changes in EEG signals.
Facial emotion recognition is an essential and important aspect of the field of human-machine interaction. Past research on facial emotion recognition focuses on the laboratory environment. However, it faces many challenges in real-world conditions, i.e., illumination changes, large pose variations and partial or full occlusions. Those challenges lead to different face areas with different degrees of sharpness and completeness. Inspired by this fact, we focus on the authenticity of predictions generated by different <emotion, region> pairs. For example, if only the mouth areas are available and the emotion classifier predicts happiness, then there is a question of how to judge the authenticity of predictions. This problem can be converted into the contribution of different face areas to different emotions. In this paper, we divide the whole face into six areas: nose areas, mouth areas, eyes areas, nose to mouth areas, nose to eyes areas and mouth to eyes areas. To obtain more convincing results, our experiments are conducted on three different databases: facial expression recognition + ( FER+), real-world affective faces database (RAF-DB) and expression in-the-wild (ExpW) dataset. Through analysis of the classification accuracy, the confusion matrix and the class activation map (CAM), we can establish convincing results. To sum up, the contributions of this paper lie in two areas: 1) We visualize concerned areas of human faces in emotion recognition; 2) We analyze the contribution of different face areas to different emotions in real-world conditions through experimental analysis. Our findings can be combined with findings in psychology to promote the understanding of emotional expressions.
This paper proposes an image encryption algorithm LQBPNN (logistic quantum and back propagation neural network) based on chaotic sequences incorporating quantum keys. Firstly, the improved one-dimensional logistic chaotic sequence is used as the basic key sequence. After the quantum key is introduced, the quantum key is incorporated into the chaotic sequence by nonlinear operation. Then the pixel confused process is completed by the neural network. Finally, two sets of different mixed secret key sequences are used to perform two rounds of diffusion encryption on the confusing image. The experimental results show that the randomness and uniformity of the key sequence are effectively enhanced. The algorithm has a secret key space greater than 2182. The adjacent pixel correlation of the encrypted image is close to 0, and the information entropy is close to 8. The ciphertext image can resist several common attacks such as typical attacks, statistical analysis attacks and differential attacks.
The aim of this work is to model and analyze the behavior of a new smart nano force sensor. To do so, the carbon nanotube has been used as a suspended gate of a metal-oxide-semiconductor field-effect transistor (MOSFET). The variation of the applied force on the carbon nanotube (CNT) generates a variation of the capacity of the transistor oxide-gate and therefore the variation of the threshold voltage, which allows the MOSFET to become a capacitive nano force sensor. The sensitivity of the nano force sensor can reach 0.124 31 V/nN. This sensitivity is greater than results in the literature. We have found through this study that the response of the sensor depends strongly on the geometric and physical parameters of the CNT. From the results obtained in this study, the increase in the applied force has as a consequence an increase in the value of the threshold voltage VTh of the MOSFET. In this paper, we first used artificial neural networks to faithfully reproduce the response of the nano force sensor model. This neural model is called direct model. Then, secondly, we designed an inverse model called an intelligent sensor which allows linearization of the response of our developed force sensor.
Energy Efficient Scheduler of Aperiodic jobs for Real-Time Embedded Systems
Driven by the ever increasing demand in function integration, more and more next generation high value-added products, such as head-up displays, solar concentrators and intra-ocular-lens, etc., are designed to possess freeform (i.e., non-rotational symmetric) surfaces. The toolpath, composed of high density of short linear and circular segments, is generally used in computer numerical control (CNC) systems to machine those products. However, the discontinuity between toolpath segments leads to high-frequency fluctuation of feedrate and acceleration, which will decrease the machining efficiency and product surface finish. Driven by the ever-increasing need for high-speed high-precision machining of those products, many novel toolpath interpolation and smoothing approaches have been proposed in both academia and industry, aiming to alleviate the issues caused by the conventional toolpath representation and interpolation methods. This paper provides a comprehensive review of the state-of-the-art toolpath interpolation and smoothing approaches with systematic classifications. The advantages and disadvantages of these approaches are discussed. Possible future research directions are also offered.
Fast high-precision patient-specific vascular tissue and geometric structure reconstruction is an essential task for vascular tissue engineering and computer-aided minimally invasive vascular disease diagnosis and surgery. In this paper, we present an effective vascular geometry reconstruction technique by representing a highly complicated geometric structure of a vascular system as an implicit function. By implicit geometric modelling, we are able to reduce the complexity and level of difficulty of this geometric reconstruction task and turn it into a parallel process of reconstructing a set of simple short tubular-like vascular sections, thanks to the easy-blending nature of implicit geometries on combining implicitly modelled geometric forms. The basic idea behind our technique is to consider this extremely difficult task as a process of team exploration of an unknown environment like a cave. Based on this idea, we developed a parallel vascular modelling technique, called Skeleton Marching, for fast vascular geometric reconstruction. With the proposed technique, we first extract the vascular skeleton system from a given volumetric medical image. A set of sub-regions of a volumetric image containing a vascular segment is then identified by marching along the extracted skeleton tree. A localised segmentation method is then applied to each of these sub-image blocks to extract a point cloud from the surface of the short simple blood vessel segment contained in the image block. These small point clouds are then fitted with a set of implicit surfaces in a parallel manner. A high-precision geometric vascular tree is then reconstructed by blending together these simple tubular-shaped implicit surfaces using the shape-preserving blending operations. Experimental results show the time required for reconstructing a vascular system can be greatly reduced by the proposed parallel technique.
Transfer Hierarchical Attention Network for Generative Dialog System
Xiang Zhang, Qiang Yang
A Survey on 3D Visual Tracking of Multicopters
Qiang Fu, Xiang-Yang Chen, Wei He
Toolpath Interpolation and Smoothing for Computer Numerical Control Machining of Freeform Surfaces: A Review