Deep Reinforcement Learning for Autonomous Driving with an Auxiliary Actor Discriminator.

In the research of robot systems, path planning and obstacle avoidance are important research directions, especially in unknown dynamic environments where flexibility and rapid decision makings are required. In this paper, a state attention network (SAN) was developed to extract features to represent the interaction between an intelligent robot and its obstacles. An auxiliary actor discriminator (AAD) was developed to calculate the probability of a collision. Goal-directed and gap-based navigation strategies were proposed to guide robotic exploration. The proposed policy was trained through simulated scenarios and updated by the Soft Actor-Critic (SAC) algorithm. The robot executed the action depending on the AAD output. Heuristic knowledge (HK) was developed to prevent blind exploration of the robot. Compared to other methods, adopting our approach in robot systems can help robots converge towards an optimal action strategy. Furthermore, it enables them to explore paths in unknown environments with fewer moving steps (showing a decrease of 33.9%) and achieve higher average rewards (showning an increase of 29.15%).

Few-Electrode EEG from the Wearable Devices Using Domain Adaptation for Depression Detection.

Nowadays, major depressive disorder (MDD) has become a crucial mental disease that endangers human health. Good results have been achieved by electroencephalogram (EEG) signals in the detection of depression. However, EEG signals are time-varying, and the distributions of the different subjects' data are non-uniform, which poses a bad influence on depression detection. In this paper, the deep learning method with domain adaptation is applied to detect depression based on EEG signals. Firstly, the EEG signals are preprocessed and then transformed into pictures by two methods: the first one is to present the three channels of EEG separately in the same image, and the second one is the RGB synthesis of the three channels of EEG. Finally, the training and prediction are performed in the domain adaptation model. The results indicate that the domain adaptation model can effectively extract EEG features and obtain an average accuracy of 77.0 ± 9.7%. This paper proves that the domain adaptation method can effectively weaken the inherent differences of EEG signals, making the diagnosis of different users more accurate.

SAFE and SBSE combined with GC-MS and GC-O for characterization of flavor compounds in Zhizhonghe Wujiapi medicinal liquor.

Volatile compounds in Chinese Zhizhonghe Wujiapi (WJP) medicinal liquor were extracted by solvent-assisted flavor evaporation extraction (SAFE) and stir bar sorptive extraction (SBSE), respectively, and identified by gas chromatography-mass spectrometry. Results showed that a total of 123 volatile compounds (i.e., 108 by SAFE, 50 by SBSE, and 34 by both) including esters, alcohols, acids, aldehydes, ketones, heterocycles, terpenes and terpenoids, alkenes, phenols, and other compounds were identified, and 67 of them were confirmed as aroma-active compounds by the application of the aroma extract dilution analysis coupled with gas chromatography-olfactometry. After making a simulated reconstitute by mixing 41 characterized aroma-active compounds (odor activity values ≥1) based on their concentrations, the aroma profile of the reconstitute showed good similarity to that of the original WJP liquor. Omission test further corroborated 34 key aroma-active compounds in the WJP liquor. The study of WJP liquor is expected to provide some insights into the characterization of special volatile components in traditional Chinese medicine liquors for the purpose of quality improvement and aroma optimization.

A Hybrid Prognostic Method for Proton-Exchange-Membrane Fuel Cell with Decomposition Forecasting Framework Based on AEKF and LSTM.

Durability and reliability are the major bottlenecks of the proton-exchange-membrane fuel cell (PEMFC) for large-scale commercial deployment. With the help of prognostic approaches, we can reduce its maintenance cost and maximize its lifetime. This paper proposes a hybrid prognostic method for PEMFCs based on a decomposition forecasting framework. Firstly, the original voltage data is decomposed into the calendar aging part and the reversible aging part based on locally weighted regression (LOESS). Then, we apply an adaptive extended Kalman filter (AEKF) and long short-term memory (LSTM) neural network to predict those two components, respectively. Three-dimensional aging factors are introduced in the physical aging model to capture the overall aging trend better. We utilize the automatic machine-learning method based on the genetic algorithm to train the LSTM model more efficiently and improve prediction accuracy. The aging voltage is derived from the sum of the two predicted voltage components, and we can further realize the remaining useful life estimation. Experimental results show that the proposed hybrid prognostic method can realize an accurate long-term voltage-degradation prediction and outperform the single model-based method or data-based method.

HS-SPME and SDE combined with GC-MS and GC-O for characterization of flavor compounds in Zhizhonghe Wujiapi medicinal liquor.

Volatile compounds in Chinese medicinal liquor, Zhizhonghe Wujiapi (WJP liquor), were extracted by headspace-solid-phase microextraction (HS-SPME) and simultaneous distillation and extraction (SDE), respectively, and identified and quantified by gas chromatography-mass spectrometry (GC-MS) and gas chromatography-olfactometry (GC-O). Results showed that a total of 133 volatile compounds (i.e., 99 by HS-SPME, 67 by SDE, and 33 by both) including esters, alcohols, acids, aldehydes, ketones, furans, terpenes, and other miscellaneous compounds were identified by GC-MS. A total of 66 aroma active compounds were further recognized by GC-O, and 43 of them were confirmed as key aroma compounds owing to their high OAV values. After making a simulated reconstitute by mixing 31 characterized aroma compounds (OAVs ≥ 1) based on their measured concentrations, the aroma profile of the reconstitute showed a good similarity to the aroma of the original WJP liquor. Omission test further corroborated 25 key aroma-active compounds in the WJP liquor. The analysis of the volatile components of this special Chinese medicinal liquor is expected to provide some insights in terms of its quality improvement and aroma profile optimization.

LandmarkNet: a 2D digital radiograph landmark estimator for registration.

BACKGROUND: Radiation therapy requires precision to target and escalate the doses to affected regions while reducing the adjacent normal tissue exposed to high radiotherapy doses. Image guidance has become the start of the art in the treating process. Registering the digital radiographs megavoltage x ray (MV-DRs) and the kilovoltage digital reconstructed radiographs (KV-DRRs) is difficult because of the poor quality of MV-DRs. We simplify the problem by registering between landmarks instead of entire image information, thence we propose a model to estimate the landmark accurately. METHODS: After doctors' analysis, it is proved that it is effective to register through several physiological features such as spinous process, tracheal bifurcation, Louis angle. We propose the LandmarkNet, a novel keypoint estimation architecture, can automatically detect keypoints in blurred medical images. The method applies the idea of Feature Pyramid Network (FPN) twice to merge the cross-scale and cross-layer features for feature extraction and landmark estimation successively. Intermediate supervision is used at the end of the first FPN to ensure that the underlying parameters are updated normally. The network finally produces heatmap to display the approximate location of landmarks and we obtain accurate position estimation after non-maximum suppression (NMS) processing. RESULTS: Our method could obtain accurate landmark estimation in the dataset provided by several cancer hospitals and labeled by ourselves. The standard percentage of correct keypoints (PCK) within 8 pixels of estimation for the spinous process, tracheal bifurcation and Louis angle is 81.24%, 98.95% and 85.61% respectively. For the above three landmarks, the mean deviation between the predicted location of each landmark and corresponding ground truth is 2.38, 0.98 and 2.64 pixels respectively. CONCLUSION: Landmark estimation based on LandmarkNet has high accuracy for different kinds of landmarks. Our model estimates the location of tracheal bifurcation especially accurately because of its obvious features. For the spinous process, our model performs well in quantity estimation as well as in position estimation. The wide application of our method assists doctors in image-guided radiotherapy (IGRT) and provides the possibility of precise treatment in the true sense.

A modality conversion approach to MV-DRs and KV-DRRs registration using information bottlenecked conditional generative adversarial network.

PURPOSE: As affordable equipment, electronic portal imaging devices (EPIDs) are wildly used in radiation therapy departments to verify patients' positions for accurate radiotherapy. However, these devices tend to produce visually ambiguous and low-contrast planar digital radiographs under megavoltage x ray (MV-DRs), which poses a tremendous challenge for clinicians to perform multimodal registration between the MV-DRs and the kilovoltage digital reconstructed radiographs (KV-DRRs) developed from the planning computed tomography. Furthermore, the existent of strong appearance variations also makes accurate registration beyond the reach of current automatic algorithms. METHODS: We propose a novel modality conversion approach to this task that first synthesizes KV images from MV-DRs, and then registers the synthesized and real KV-DRRs. We focus on the synthesis technique and develop a conditional generative adversarial network with information bottleneck extension (IB-cGAN) that takes MV-DRs and nonaligned KV-DRRs as inputs and outputs synthesized KV images. IB-cGAN is designed to address two main challenges in deep-learning-based synthesis: (a) training with a roughly aligned dataset suffering from noisy correspondences; (b) making synthesized images have real clinical meanings that faithfully reflects MV-DRs rather than nonaligned KV-DRRs. Accordingly, IB-cGAN employs (a) an adversarial loss to provide training supervision at semantic level rather than the imprecise pixel level; (b) an IB to constrain the information from the nonaligned KV-DRRs. RESULTS: We collected 2698 patient scans to train the model and 208 scans to test its performance. The qualitative results demonstrate realistic KV images can be synthesized allowing clinicians to perform the visual registration. The quantitative results show it significantly outperforms current nonmodality conversion methods by 22.37% (P = 0.0401) in terms of registration accuracy. CONCLUSIONS: The modality conversion approach facilitates the downstream MV-KV registration for both clinicians and off-the-shelf registration algorithms. With this approach, it is possible to benefit the developing countries where inexpensive EPIDs are widely used for the image-guided radiation therapy.

A model of chronic, transmissible Otitis Media in mice.

Infection and inflammation of the middle ears that characterizes acute and chronic otitis media (OM), is a major reason for doctor visits and antibiotic prescription, particularly among children. Nasopharyngeal pathogens that are commonly associated with OM in humans do not naturally colonize the middle ears of rodents, and experimental models in most cases involve directly injecting large numbers of human pathogens into the middle ear bullae of rodents, where they induce a short-lived acute inflammation but fail to persist. Here we report that Bordetella pseudohinzii, a respiratory pathogen of mice, naturally, efficiently and rapidly ascends the eustachian tubes to colonize the middle ears, causing acute and chronic histopathological changes with progressive decrease in hearing acuity that closely mimics otitis media in humans. Laboratory mice experimentally inoculated intranasally with very low numbers of bacteria consistently have their middle ears colonized and subsequently transmit the bacterium to cage mates. Taking advantage of the specifically engineered and well characterized immune deficiencies available in mice we conducted experiments to uncover different roles of T and B cells in controlling bacterial numbers in the middle ear during chronic OM. The iconic mouse model provides significant advantages for elucidating aspects of host-pathogen interactions in otitis media that are currently not possible using other animal models. This natural model of otitis media permits the study of transmission between hosts, efficient early colonization of the respiratory tract, ascension of the eustachian tube, as well as colonization, pathogenesis and persistence in the middle ear. It also allows the combination of the powerful tools of mouse molecular immunology and bacterial genetics to determine the mechanistic basis for these important processes.

[Rapid on-site screening of five prohibited ingredients in cosmetics using thermal desorption-corona discharge ionization coupled with ion mobility spectrometry].

A rapid on-site analytical method is developed for screening lidocaine and four other types of prohibited ingredients in cosmetics using thermal desorption-corona discharge ionization together with ion mobility spectrometry. Performing no pretreatment, we dropped or sprayed cosmetic samples onto a Nomex sampling swab. The sampler was placed into a compartment for thermal desorption and corona discharge ionization; then, ion mobility spectrometry analysis was performed. The limits of detection (LODs) for the five analytes ranged from 10 to 50 ng. The instrumental analysis time for a single run was less than 20 ms, and the total sample analysis period was within 1 min. The proposed method is simple, fast, and has low cost, and could be used as an analytical tool for rapid on-site screening of prohibited ingredients in cosmetics.

Quantized Feedback Control of Fuzzy Markov Jump Systems.

This paper addresses the problem of quantized feedback control of nonlinear Markov jump systems (MJSs). The nonlinear plant is represented by a class of fuzzy MJSs with time-varying delay based on a Takagi-Sugeno fuzzy model. The quantized signal is utilized for control purpose and the sector bound approach is exploited to deal with quantization errors. By constructing a Lyapunov function which depends both on mode information and fuzzy basis functions, the reciprocally convex approach is used to derive the criterion which is able to ensure the stochastic stability with a predefined l2-l∞ performance of the resulting closed-loop system. The design of the quantized feedback controller is then converted to a convex optimization problem, which can be handled through the linear matrix inequality technique. Finally, a simulation example is presented to verify the effectiveness and practicability of the proposed new design techniques.

Improved linear quadratic and proportional control system for improved liquid level system regulation in a coke fractionation tower.

A new linear quadratic regulation (LQ) control plus a proportional (P) control system is proposed for the level regulation in an industrial coke fractionation tower. The process is first stabilized using a P controller and then a subsequent LQ controller is designed for the P control system. The P control system is modeled as a generalized first order plus dead time (FOPDT) process using step-response test and the LQ-P controller is designed through a new state space structure. Performance in terms of regulatory and servo issues were investigated. Simulation results showed that the proposed method is more robust and improves performance than traditional model predictive control.

Improved control using extended non-minimal state space MPC and modified LQR for a kind of nonlinear systems.

Inspired by the state space model based predictive control, this paper presents the combination design of extended non-minimal state space predictive control (ENMSSPC) and modified linear quadratic regulator (LQR) for a kind of nonlinear process with output feedback coupling, which shows improved control performance for both model/plant match and model/plant mismatch cases. In many previous control methods for this kind of nonlinear systems, the nonlinear part is treated in different ways such as ignored, represented as a rough linear one or assumed to be time-variant when corresponding predictive control methods are designed. However, the above methods will generally lead to information loss, resulting in the influenced control performance. This paper will show that the ENMSSPC-LQ control structure will further improve closed-loop control performance concerning tracking ability and disturbance rejection compared with previous predictive control methods.

Online soft sensor of humidity in PEM fuel cell based on dynamic partial least squares.

Online monitoring humidity in the proton exchange membrane (PEM) fuel cell is an important issue in maintaining proper membrane humidity. The cost and size of existing sensors for monitoring humidity are prohibitive for online measurements. Online prediction of humidity using readily available measured data would be beneficial to water management. In this paper, a novel soft sensor method based on dynamic partial least squares (DPLS) regression is proposed and applied to humidity prediction in PEM fuel cell. In order to obtain data of humidity and test the feasibility of the proposed DPLS-based soft sensor a hardware-in-the-loop (HIL) test system is constructed. The time lag of the DPLS-based soft sensor is selected as 30 by comparing the root-mean-square error in different time lag. The performance of the proposed DPLS-based soft sensor is demonstrated by experimental results.

Field programmable gate array based parallel strapdown algorithm design for strapdown inertial navigation systems.

A new generalized optimum strapdown algorithm with coning and sculling compensation is presented, in which the position, velocity and attitude updating operations are carried out based on the single-speed structure in which all computations are executed at a single updating rate that is sufficiently high to accurately account for high frequency angular rate and acceleration rectification effects. Different from existing algorithms, the updating rates of the coning and sculling compensations are unrelated with the number of the gyro incremental angle samples and the number of the accelerometer incremental velocity samples. When the output sampling rate of inertial sensors remains constant, this algorithm allows increasing the updating rate of the coning and sculling compensation, yet with more numbers of gyro incremental angle and accelerometer incremental velocity in order to improve the accuracy of system. Then, in order to implement the new strapdown algorithm in a single FPGA chip, the parallelization of the algorithm is designed and its computational complexity is analyzed. The performance of the proposed parallel strapdown algorithm is tested on the Xilinx ISE 12.3 software platform and the FPGA device XC6VLX550T hardware platform on the basis of some fighter data. It is shown that this parallel strapdown algorithm on the FPGA platform can greatly decrease the execution time of algorithm to meet the real-time and high precision requirements of system on the high dynamic environment, relative to the existing implemented on the DSP platform.

Integrated Design and Implementation of Embedded Control Systems with Scilab.

Embedded systems are playing an increasingly important role in control engineering. Despite their popularity, embedded systems are generally subject to resource constraints and it is therefore difficult to build complex control systems on embedded platforms. Traditionally, the design and implementation of control systems are often separated, which causes the development of embedded control systems to be highly timeconsuming and costly. To address these problems, this paper presents a low-cost, reusable, reconfigurable platform that enables integrated design and implementation of embedded control systems. To minimize the cost, free and open source software packages such as Linux and Scilab are used. Scilab is ported to the embedded ARM-Linux system. The drivers for interfacing Scilab with several communication protocols including serial, Ethernet, and Modbus are developed. Experiments are conducted to test the developed embedded platform. The use of Scilab enables implementation of complex control algorithms on embedded platforms. With the developed platform, it is possible to perform all phases of the development cycle of embedded control systems in a unified environment, thus facilitating the reduction of development time and cost.

Cross-Layer Adaptive Feedback Scheduling of Wireless Control Systems.

There is a trend towards using wireless technologies in networked control systems. However, the adverse properties of the radio channels make it difficult to design and implement control systems in wireless environments. To attack the uncertainty in available communication resources in wireless control systems closed over WLAN, a cross-layer adaptive feedback scheduling (CLAFS) scheme is developed, which takes advantage of the co-design of control and wireless communications. By exploiting crosslayer design, CLAFS adjusts the sampling periods of control systems at the application layer based on information about deadline miss ratio and transmission rate from the physical layer. Within the framework of feedback scheduling, the control performance is maximized through controlling the deadline miss ratio. Key design parameters of the feedback scheduler are adapted to dynamic changes in the channel condition. An eventdriven invocation mechanism for the feedback scheduler is also developed. Simulation results show that the proposed approach is efficient in dealing with channel capacity variations and noise interference, thus providing an enabling technology for control over WLAN.

Diagnosis of gastric cancer using decision tree classification of mass spectral data.

Although gastric cancer is the second leading cause of cancer death worldwide, specific and sensitive biomarkers that can be used for its diagnosis are still unavailable. Attempting to improve on current approaches to the serological diagnosis of gastric cancer, we subjected serum samples from 245 individuals (including 127 gastric cancer patients, 100 age- and sex-matched healthy individuals, nine benign gastric lesion patients and nine colorectal cancer patients) for analysis by surface-enhanced laser desorption/ionization (SELDI) mass spectrometry. Peaks were detected with Ciphergen SELDI software version 3.1.1 and analyzed with Biomarker Patterns' software 5.0. We developed a classifier for separating the gastric cancer groups from the healthy groups. Three protein masses with 1468, 3935 and 7560 m/z were selected as a potential 'fingerprint' for the detection of gastric cancer. It was able to distinguish the gastric cancer patients from the health volunteers with a sensitivity of 95.6% and a specificity of 92.0% in the training set. In the blinding set, it was capable of differentiating the gastric cancer samples from the others with a specificity of 88.0%, a sensitivity of 85.3%, and an accuracy of 86.4%. These values were all higher than those achieved in a parallel analysis by measuring serum carcinoembryonic antigen (CEA) and carbohydrate antigen (CA)19-9 together. Therefore, the decision tree analysis of serum proteomic patterns has the potential to be used in gastric cancer diagnosis.

[Design and control of in vitro pulse modulated microwave exposure system].

This paper presents the design and development of a set of microwave exposure system based on 1.8GHz mobile RF signal. This system can work on several modulation types to do microwave exposure experiment under different specific absorption rate (SAR) and prepare the way for researches in the effect exerted by the electromagnetic signal of mobile on human health. The hardware is made up of several RF instruments, waveguide and computer, and the software introduces the accomplishment of the control system and the algorithm of control.