Search

Although traffic conflict techniques have proven to be effective means for road safety analysis, they still suffer from incomplete conceptualization, observer subjectivity, and high data collection cost. To address these problems, video analysis has been increasingly applied to gain a better understanding of the behaviors of road users based on detailed motion data. However, the motion patterns underlying these data are rarely extracted to study the safety of their interactions. This article presents a vision-based method of traffic conflict detection through learning motion patterns from trajectories, for which an original algorithm was established through clustering and subsequent modeling. Using the extracted path and velocity information, we clustered trajectories hierarchically by applying an improved fuzzy $K$ -means algorithm with a modified Hausdorff distance. Each obtained cluster was taken as a labeled set to determine the structure and train the parameters of a hidden Markov model (HMM) that encoded the spatiotemporal characteristics of the trajectories as motion patterns. Based on the targeted trajectory predictions by the learned HMMs following the conflict development, a probabilistic model was developed to estimate the collision likelihood between vehicles to identify traffic conflicts. The experimental results obtained using actual traffic videos demonstrated the applicability of the algorithms for learning motion patterns and the feasibility of the approach for traffic conflict detection. The predicted trajectories were sufficiently accurate to calculate the collision probability, which was qualified as an indicator for measuring the conflict severity. These findings will have important implications for effective improvements in active road safety.

Speech feature learning is the key to speech mental health recognition. Deep feature learning can automatically extract the speech features but suffers from the small sample problem. The traditional feature extract method is effective, but cannot find the inter-feature structure to generate the new high-quality features. This paper proposes an embedded hybrid feature deep sparse stacked autoencoder ensemble method to solve this problem. Firstly, the speech features are extracted based on prior knowledge and called original features. Secondly, the original features are embedded into the deep network (Sparse Stacked Autoencoder) to filter the output of the hidden layer, to enhance the complementarity between the deep features and the original features. Thirdly, the L1 regularized feature selection mechanism is designed to reduce the hybrid feature set formed by the combination of deep features and original features. Finally, a manifold projection classifier ensemble is designed to enhance the stability of classification. Besides, this paper for the first time proposes a speech collection scheme for mental health recognition. We construct a large-scale Chinese mental health speech database for verification of the proposed algorithm of mental health. In the experimental section, the proposed algorithm is verified and compared with the representative related algorithms. The experimental results show that the proposed algorithm has better classification accuracy than the other representative algorithms. The proposed method combines the advantages of deep feature learning and traditional feature extraction methods more efficiently to solve the small sample problem.

Accurately predicting the trajectory of pedestrians helps autonomous vehicles to drive safely. In this paper, a work of predicting the trajectory of pedestrians by considering their posture is described. Two seconds of historical data are used to predict the pedestrian's actions in the next second based on a long short-term memory approach. The purpose of this experiment is to estimate whether pedestrians will cross the road in a mid-block setting without crosswalks and what paths they will take. The scene of this experiment is located on a street near the campus of the University of California at Berkeley.

In this survey, we systematically summarize the current literature on studies that apply reinforcement learning (RL) to the motion planning and control of autonomous vehicles. Many existing contributions can be attributed to the pipeline approach, which consists of many hand-crafted modules, each with a functionality selected for the ease of human interpretation. However, this approach does not automatically guarantee maximal performance due to the lack of a system-level optimization. Therefore, this paper also presents a growing trend of work that falls into the end-to-end approach, which typically offers better performance and smaller system scales. However, their performance also suffers from the lack of expert data and generalization issues. Finally, the remaining challenges applying deep RL algorithms on autonomous driving are summarized, and future research directions are also presented to tackle these challenges., Comment: 8 pages

Learning from demonstrations has made great progress over the past few years. However, it is generally data hungry and task specific. In other words, it requires a large amount of data to train a decent model on a particular task, and the model often fails to generalize to new tasks that have a different distribution. In practice, demonstrations from new tasks will be continuously observed and the data might be unlabeled or only partially labeled. Therefore, it is desirable for the trained model to adapt to new tasks that have limited data samples available. In this work, we build an adaptable imitation learning model based on the integration of Meta-learning and Adversarial Inverse Reinforcement Learning (Meta-AIRL). We exploit the adversarial learning and inverse reinforcement learning mechanisms to learn policies and reward functions simultaneously from available training tasks and then adapt them to new tasks with the meta-learning framework. Simulation results show that the adapted policy trained with Meta-AIRL can effectively learn from limited number of demonstrations, and quickly reach the performance comparable to that of the experts on unseen tasks., Comment: 2021 International Conference on Robotics and Automation (ICRA 2021)

The increasing penetration of distributed generation transforms the traditional passive distribution network into an active network, which will result in the flow of fault current changing from unidirectional to bidirectional, and the traditional single-source radial distribution network fault location methods are no longer applicable. In this paper, a fault location method for active distribution network with distributed generation is proposed. First, the correlation matrix corresponding to each power source acting individually is established according to the line topology, and a switching function that reflects the relationship between the fault state of a line segment and the FTU alarm signals is constructed using this matrix as a bridge. Then, the objective function reflecting the similarity between the desired signals and the actual alarm signals is established by combining the minimum set theory, and the fault probability of each segment in the line is calculated by using the BPSO algorithm. Finally, an active distribution network containing 13 nodes is used as an example for simulation and verification. The simulation results not only validate the effectiveness of the proposed method in this paper in the case of single-point fault and double faults, but also prove to be highly fault-tolerant.

Bearing plays decisive roles in modern industrial and electrical foundations. For authentic situation, immensely streaming and distributed data are congregated by Prognostics and Health Management (PHM) systems. The massive rigid data conduces the following puzzle: comparable huge excesses for PHM system, which is bounded on the whole huge sets. For this task, we employ active learning framework. In this paper, we firstly propose a novel nonparametric active learning (NAL) method and prove that NAL acquisition function is a tightly upper-bound of naive form. We validate our method on TCN (Temporal Convolutional Network) and achieve the state of the art performance on CWRU benchmark, providing mighty data effectiveness enhancement on industrial field.

This paper presents the first 802.11ax compliant iterative detection and decoding (IDD) receiver that supports up to 4 $\times$ 4 1024-QAM MIMO detection in the open literature. Soft-input-soft-output (SISO) MIMO detection is implemented with a lattice reduction aided (LRA) K-best searcher and a max-log list demapper. A hardware-efficient IDD receiver is proposed to achieve the required packet-rate (PER) with a feasible latency. The extrinsic information transfer (EXIT) chart is utilized to reduce the number of iterations for IDD. Given the 802.11ax latency constraint, the performance, power, area (PPA) design space is explored to identify the optimal IDD receiver architecture. 50% of IDD inner iterations are reduced with only a 0.05dB loss in PER. The proposed IDD receiver achieves a 1dB improvement in PER with 3.6$\times$ smaller area and 3.0 $\times$ lower power consumption when compared to the best non-IDD receiver.

Lane-change maneuvers are commonly executed by drivers to follow a certain routing plan, overtake a slower vehicle, adapt to a merging lane ahead, etc. However, improper lane change behaviors can be a major cause of traffic flow disruptions and even crashes. While many rule-based methods have been proposed to solve lane change problems for autonomous driving, they tend to exhibit limited performance due to the uncertainty and complexity of the driving environment. Machine learning-based methods offer an alternative approach, as Deep reinforcement learning (DRL) has shown promising success in many application domains including robotic manipulation, navigation, and playing video games. However, applying DRL to autonomous driving still faces many practical challenges in terms of slow learning rates, sample inefficiency, and safety concerns. In this study, we propose an automated lane change strategy using proximal policy optimization-based deep reinforcement learning, which shows great advantages in learning efficiency while still maintaining stable performance. The trained agent is able to learn a smooth, safe, and efficient driving policy to make lane-change decisions (i.e. when and how) in a challenging situation such as dense traffic scenarios. The effectiveness of the proposed policy is validated by using metrics of task success rate and collision rate. The simulation results demonstrate the lane change maneuvers can be efficiently learned and executed in a safe, smooth, and efficient manner.

Reliable representation of multi-vehicle interactions in urban traffic is pivotal but challenging for autonomous vehicles due to the volatility of the traffic environment, such as roundabouts and intersections. This paper describes a semi-stochastic potential field approach to represent multi-vehicle interactions by integrating a deterministic field approach with a stochastic one. First, we conduct a comprehensive evaluation of potential fields for representing multi-agent intersections from the deterministic and stochastic perspectives. For the former, the estimates at each location in the region of interest (ROI) are deterministic, which is usually built using a family of parameterized exponential functions directly. For the latter, the estimates are stochastic and specified by a random variable, which is usually built based on stochastic processes such as the Gaussian process. Our proposed semi-stochastic potential field, combining the best of both, is validated based on the INTERACTION dataset collected in complicated real-world urban settings, including intersections and roundabout. Results demonstrate that our approach can capture more valuable information than either the deterministic or stochastic ones alone. This work sheds light on the development of algorithms in decision-making, path/motion planning, and navigation for autonomous vehicles in the cluttered urban settings.

An integrated design framework is proposed to automate radiation-harden (rad-hard) VLSI systems in a standard CMOS technology. TMR, DICE, SERL, ELT, and ECC techniques are integrated across architecture, circuit, and layout levels. Performance of the rad-hard cells were evaluated in 0.18m CMOS. A rad-hard RISC processor targeting for an inclination micro-satellite on a 720km orbit was realized in 90nm CMOS. The chip was tested by applying heavy ions with corresponding radiation dose. The rad-hard RISC processor functions under all the test conditions (LET $ \lt 101.5$ MeV-cm $^{2} /$ mg), validating the effectiveness of the methodology.

Thermal-mechanical coupling damage is an important factor leading to instability and failure of rock mass engineering. The constitutive relation and damage evolution of rock-like materials under high strain rate is crucial parameters for the design of underground rock mass structures, and is also an essential foundation for the analysis of instability phenomena in rock mass engineering. In this paper, a series of physical properties, static, and dynamic mechanical tests were carried out on Changsha sandstone after different high temperature treatments (ranging from 25 °C to 800 °C). Results indicate that thermal treatment effectively weakens the sandstone specimens. Under dynamic loading, the rock shows obvious strain rate effect, and the stress-strain curve changes from brittleness to viscoelasticity with the increase of temperature. Based on statistical damage theory and Weibull distribution, combining the analysis of the change laws of stress-strain curves of sandstone with temperature, a damage constitutive model that can reflect the variation in dynamic mechanical properties with temperature was proposed, for which the ultrasonic velocity is used to characterize the initial damage of rock caused by high temperature. Thereafter, the rationality of the constitutive model was verified by experiments. Combing damage evolution characteristics and rock failure process recorded by high-speed camera, the dynamic stress-strain relation curves of rock under different high temperature are divided into four stages, and the damage evolution process of rock under thermo-mechanical coupling was also analyzed. Results indicate that the values of thermo-mechanical damage increase exponentially with rock strain, and the damage evolution rate presents a two stage variation of first increase and then decrease. The initial damage stress of rock decreases with increasing temperature, but always maintains a stable percentage compared with the peak stress of the rock. Although the damage value at the peak point of stress increases with the increase of the temperature, it did not reach the maximum value of 1, which indicates that the rock damage continues to increase in the post peak loading stage. The findings of this paper can provide guidance for the macroscopic mechanical damage of rock under high temperatures and dynamic loading.

To learn a reward function that a driver adheres to is of importance to the human-like design of autonomous driving systems. Inverse reinforcement learning (IRL) is one of the recent advances that can achieve this objective, but it often suffers from the low efficiency of generating optimal policy by reinforcement learning (RL) each time when updating reward weights. This paper presents an accelerated IRL method by approaching the optimal policy among randomly pre-sampled policies in designed sub-space instead of finding it through RL in the whole policy space. The corresponding trajectories are targeted via an optimal trajectory selector in the candidate trajectory library generated by pre-sampled policies. The weights then are updated by comparing the selected trajectories and the expert ones. The proposed method is very suitable for improving learning efficiency for low-dimensional problems like autonomous driving, whose expert policies are nearly tractable. Results with simulated driving data show that it only took 11 iterations to converge, while the average longitudinal RMS error of the recovered trajectories based on the learned reward function was only 2.14m.

To fulfill high-level automation, an automated vehicle needs to learn to make decisions and control its movement under complex scenarios. Due to the uncertainty and complexity of the driving environment, most classical rule-based methods cannot solve the problem of complicated decision tasks. Deep reinforcement learning has demonstrated impressive achievements in many fields such as playing games and robotics. However, a direct application of reinforcement learning algorithm for automated driving still face challenges in handling complex driving tasks. In this paper, we proposed a hierarchical reinforcement learning based architecture for decision making and control of lane changing situations. We divided the decision and control process into two correlated processes: 1) when to conduct lane change maneuver and 2) how to conduct the maneuver. To be specific, we first apply Deep Q-network (DQN) to decide when to conduct the maneuver based on the consideration of safety. Subsequently, we design a Deep Q-learning framework with quadratic approximator for deciding how to complete the maneuver in longitudinal direction (e.g. adjust to the selected gap or just follow the preceding vehicle). Finally, a polynomial lane change trajectory is generated and Pure Pursuit Control is implemented for path tracking for the lane change situation. We demonstrate the effectiveness of this framework in simulation, from both the decision-making and control layers.

In this paper, a switching strategy is proposed for the extended phase shift (EPS) controlled dual active bridge (DAB) converter which using the hybrid switch (HyS) configured by paralleling SiC MOSFET and Si IGBT, considering the special soft switching characteristic of DAB. For most of the existing literature, the common switching strategy of HyS is to switch on the SiC MOSFET before Si IGBT and switch off the SiC MOSFET after Si IGBT. However, if only using this switching strategy, the SiC MOSFET may be damaged by long-time operating at the over rated current condition. The proposed switching strategy for the HyS is that the SiC MOSFET is only conducted for a short while during the process of switching Si IGBT on or off. The proposed switching strategy can shorten the conduction time of SiC MOSFET, decrease the losses of HyS and extend the safe operation area of HyS. Simulation results on a 30 kW HyS DAB converter verify the availability of the proposed switching strategy.

Lane change is a common type of driving behaviors in daily traveling, and is an essential functionality for highly automated vehicles. For the design of proper driving behaviors of automated vehicles, it is important to have a comprehensive understanding of the real-world driving behaviors. In this paper, we conduct an investigation of the lane change behaviors by examining vehicle trajectories, headways and spacing gaps between adjacent vehicles in a real-world traffic dataset, NGSIM. In particular, the identification of lane change trajectories is performed, and the distribution of headways and spacing gaps are extracted. Results show that the characteristics of lane changing actions provide new insight and serve as a valuable reference for understanding of driving behaviors in real world situations.

Lane change is a challenging task which requires delicate actions to ensure safety and comfort. Some recent studies have attempted to solve the lane-change control problem with Reinforcement Learning (RL), yet the action is confined to discrete action space. To overcome this limitation, we formulate the lane change behavior with continuous action in a model-free dynamic driving environment based on Deep Deterministic Policy Gradient (DDPG). The reward function, which is critical for learning the optimal policy, is defined by control values, position deviation status, and maneuvering time to provide the RL agent informative signals. The RL agent is trained from scratch without resorting to any prior knowledge of the environment and vehicle dynamics since they are not easy to obtain. Seven models under different hyperparameter settings are compared. A video showing the learning progress of the driving behavior is available. It demonstrates the RL vehicle agent initially runs out of road boundary frequently, but eventually has managed to smoothly and stably change to the target lane with a success rate of 100% under diverse driving situations in simulation., Published at the 30th IEEE Intelligent Vehicles Symposium (IV), 2019

A reliable controller is critical for execution of safe and smooth maneuvers of an autonomous vehicle. The controller must be robust to external disturbances, such as road surface, weather, wind conditions, and so on. It also needs to deal with internal variations of vehicle sub-systems, including powertrain inefficiency, measurement errors, time delay, etc. These factors introduce issues in controller performance. In this paper, a feed-forward compensator is designed via a data-driven method to model and optimize the controller performance. Principal Component Analysis (PCA) is applied for extracting influential features, after which a Time Delay Neural Network is adopted to predict control errors over a future time horizon. Based on the predicted error, a feedforward compensator is then designed to improve control performance. Simulation results in different scenarios show that, with the help of with the proposed feedforward compensator, the maximum path tracking error and the steering wheel angle oscillation are improved by 44.4% and 26.7%, respectively., Published at the 30th IEEE Intelligent Vehicles Symposium, 2019. arXiv admin note: substantial text overlap with arXiv:1901.11212

As autonomous vehicles (AVs) need to interact with other road users, it is of importance to comprehensively understand the dynamic traffic environment, especially the future possible trajectories of surrounding vehicles. This paper presents an algorithm for long-horizon trajectory prediction of surrounding vehicles using a dual long short term memory (LSTM) network, which is capable of effectively improving prediction accuracy in strongly interactive driving environments. In contrast to traditional approaches which require trajectory matching and manual feature selection, this method can automatically learn high-level spatial-temporal features of driver behaviors from naturalistic driving data through sequence learning. By employing two blocks of LSTMs, the proposed method feeds the sequential trajectory to the first LSTM for driver intention recognition as an intermediate indicator, which is immediately followed by a second LSTM for future trajectory prediction. Test results from real-world highway driving data show that the proposed method can, in comparison to state-of-art methods, output more accurate and reasonable estimate of different future trajectories over 5s time horizon with root mean square error (RMSE) for longitudinal and lateral prediction less than 5.77m and 0.49m, respectively., Published at the 21st International Conference on Intelligent Transportation Systems (ITSC), 2018

The spatio-temporal pattern of rice exposure to heat stress (EHS) would provide strong supports for decision makers particularly in China, to cope with increasing heat-stress threats under global warming. To obtain reliable assessments, the study period needs no less than thirty years. But how much would different temporal scales affect rice EHS pattern in China? Whether these differences potentially affect adaptation decisions? These problems still remain unclear. Here, we assessed the spatio-temporal patterns of rice EHS at three typical-used scales (1980–2009, 1970–2009 and 1960–2009), and accordingly projected EHS over the 2010s to further explore which scale could better capture recent EHS levels in China. The results showed that in the east of Sichuan Basin and the northwest of the mid-lower reaches of Yangtze River, EHS was highest over 1960–2009, whereas southern China saw highest EHS over 1980–2009. The temporal pattern saw large differences in both the areas with significant trends and their trend levels, showing particular hot spots over 1980–2009. The projected EHS based on the datasets over 1970–2009 were more consistent with those over the 2010s in most cases. Based on the patterns at different temporal scales, adaptation decisions would also show different characteristics, such as more attentions would be paid to the south of Anhui province over the scale of 1980–2009.

Lane change is a crucial vehicle maneuver which needs coordination with surrounding vehicles. Automated lane changing functions built on rule-based models may perform well under pre-defined operating conditions, but they may be prone to failure when unexpected situations are encountered. In our study, we proposed a Reinforcement Learning based approach to train the vehicle agent to learn an automated lane change behavior such that it can intelligently make a lane change under diverse and even unforeseen scenarios. Particularly, we treated both state space and action space as continuous, and designed a Q-function approximator that has a closed- form greedy policy, which contributes to the computation efficiency of our deep Q-learning algorithm. Extensive simulations are conducted for training the algorithm, and the results illustrate that the Reinforcement Learning based vehicle agent is capable of learning a smooth and efficient driving policy for lane change maneuvers., Comment: 6 pages, 4 figures, 2018 IEEE Intelligent Vehicle Symposium

This work presents a reconfigurable processor based on the alternating direction method of multipliers (ADMM) algorithm for reconstructing compressively-sensed signals. The chip delivers a throughput of 573-to-2,901KS/s for reconstructing physiological signals. It dissipates 12.6mW at 87 MHz at 0.6V. Compared to the state-of-the-art designs, the chip achieves a 5.7-to-14x higher throughput with 5-to-11x lower energy for the target reconstruction SNR (RSNR) ≥ 15dB.

Multiple automakers have in development or in production automated driving systems (ADS) that offer freeway-pilot functions. This type of ADS is typically limited to restricted-access freeways only, that is, the transition from manual to automated modes takes place only after the ramp merging process is completed manually. One major challenge to extend the automation to ramp merging is that the automated vehicle needs to incorporate and optimize long-term objectives (e.g. successful and smooth merge) when near-term actions must be safely executed. Moreover, the merging process involves interactions with other vehicles whose behaviors are sometimes hard to predict but may influence the merging vehicle optimal actions. To tackle such a complicated control problem, we propose to apply Deep Reinforcement Learning (DRL) techniques for finding an optimal driving policy by maximizing the long-term reward in an interactive environment. Specifically, we apply a Long Short-Term Memory (LSTM) architecture to model the interactive environment, from which an internal state containing historical driving information is conveyed to a Deep Q-Network (DQN). The DQN is used to approximate the Q-function, which takes the internal state as input and generates Q-values as output for action selection. With this DRL architecture, the historical impact of interactive environment on the long-term reward can be captured and taken into account for deciding the optimal control policy. The proposed architecture has the potential to be extended and applied to other autonomous driving scenarios such as driving through a complex intersection or changing lanes under varying traffic flow conditions., IEEE International Conference on Intelligent Transportation Systems, Yokohama, Japan, 2017

Most present methods of saliency detection emphasize too much on the local contrast while ignore the global feature of image. The detailed characteristics of the image can be reflected based on the local comparison of image. However, the overall saliency of the image cannot be reflected. In this paper, a saliency detection model combined local and global features was proposed. Firstly, a local feature saliency map was produced by combing background saliency map generated by multi-feature mode and foreground saliency map generated by foreground region contrast method. Then the deep convolution neural network (CNN) was used to train the global image with the center of superpixel block, the label of which is the ground truth of the center superpixel. Thus the global feature saliency map was acquired. The final saliency map was presented by merging local saliency and the global saliency map together. This approach was extensively evaluated on three public datasets including ECSSD, PASCAL-S, MSRA-5000. The higher F-measure (higher is better) as well as lower MAE value (lower is better) than conventional algorithms are obtained in this approach. In addition, the indoor environment images collected in service robot laboratory of Shandong University also obtained a good saliency detection effect through the proposed method.

In order to explore the major traffic accident factors for different age groups, this paper adopts the accident data of Statewide Integrated Traffic Records System (SWITRS) for empirical study. 33,245 accident records in a state from January 1st, 2010 to December 31st, 2011 are selected. 8 accident factors, as well as their corresponding sub-factors, are analyzed for both senior-driver group and young-driver group. A data processing method for traffic accidents based on an adaptive boosting algorithm (AdaBoost) is proposed. According to the analysis results, the major factors that influence the driving safety of the senior driver group are weather and the time of day, the weight of which are 0.28 and 0.22, respectively. In particular, the rainy and snowy days, the sub-factors from the weather variable, influence old drivers most significantly. In contrast, the main factors that affect the driving safety of young people are traffic violation and vehicle behavior, with a weight of 0.25 and 0.19, respectively. Hence, it could be inferred that the diverse manifestation patterns of traffic accidents are related to drivers' age.

Open Stack is an increasingly popular open source solutions for the deployment of a service (IaaS) cloud. As an Open Stack user or administrator, you often need to write scripts to automate common tasks. In addition to the REST and command line interfaces, Open Stack is also disclosed a native Python API bindings. Learn how to use these Python bindings simplify the process of writing automated Open Stack scripts to improve the efficiency of the administrator. Open Stack comes with a dashboard Web application which is ideal for performing manual tasks, such as starting a single virtual machine (VM) instances, however, if you want to automate cloud-based tasks, you need to write a script which is able to operate Open Stack. It is incredible to imagine back again for using the REST API or command-line tools to build your automation scripts after Learn Python API.

A logical data petri net (LDPN), a logical data workflow net (LDWN) and a collaborative logical data workflow net (CLDWN) are represented. They are improved formal models extended with data variables, guards and the output condition expressions based on our previous models. Data variables are used to represent data. Guards are used to indicate the additional constraint related to data except for a token. The output condition expressions are used to clearly show the condition in which post places of logical output transitions get a token. A typical example of collaborative electronic commerce systems with the batch function and passing value indeterminacy is used to show the advantage of the improved models.

Left-turn collisions are one of the most common types of traffic accidents at intersections because of the lack of broad vision. Collision warning system provides an effective way to solve the problem, but the poor accuracy of the trajectory prediction module in the collision warning system reduces it reliability. In this paper, we studied the trajectory prediction of left-turn vehicles at T-shaped intersections based on the technology of Location Based Service (LBS) which integrates Global Positioning System (GPS), Geographic Information System (GIS), together with sensors and telecommunication and provides real-time vehicle related information, such as position, velocity, acceleration, etc. Specifically, a Double-Kalman Filter (DKF) consisting of a yaw angle Kalman Filter and a position KalmanFitler is developed, in which the output of the yaw angle Kalman Filter is taken as one of the inputs to the position Kalman Filter to improve the prediction accuracy. Results show that the left-turn trajectory predicted by DKF is fairly close to the true trajectory with errors below 1 meter whereas raw measurements produce errors at around 3 meters. The improved accuracy of predicted trajectory plays a key role in ensuring a reliable and effective collision warning system.

Video sensors are widely used in many applications such as security monitoring and home care. However, the growth of the number of sensors makes it impractical to stream all videos back to a central server for further processing, due to communication bandwidth and server storage constraints. Multi-view video summarization allows us to discard redundant data in the video streams taken by a group of sensors. All prior multi-view summarization methods, however, process video data in an off-line and centralized manner, which means that all videos are still required to be streamed back to the server before conducting the summarization. This paper proposes an on-line, distributed multi-view summarization system, which integrates the ideas of Maximal Marginal Relevance (MMR) and MS-Wave, a bandwidth-efficient distributed algorithm for finding k-nearest-neighbors and k-farthest-neighbors. Empirical studies show that our proposed system can discard redundant videos and keep important keyframes as effectively as centralized approaches, while transmitting only 1/6 to 1/3 as much data.

To solve the problem of multi-target tracking model with the time-varying number of targets, an improved Cubature particle PHD (CP-PHD) filter is proposed for multi-target tracking system. Firstly, the improved third-degree Spherical-Radial rule is applied to calculate the probability distribution of the nonlinear stochastic function; it introduces an adaptive memory factor for generating the importance density function based on the cubature Kalman filter (CKF). Thus the desirable particles obtained avoiding the effect of the old data. Then prediction and update the random finite set of multi-target using a bank of Gaussian particle filters. The method approximates the updated PHD into the form of Gaussian mixture using the particles with maximum. The simulation results demonstrated the proposed algorithm can effectively deal with the multi-target tracking problems with non-linear non-Gaussian model. Compared with the Gaussian particle PHD filter (GP-PHDF), the proposed algorithm can reduce the multi-target distance error by nearly 70% and reduce the running time by 20%.

In M2M applications, it is very common to encounter the ad hoc snapshot query that requires fast responses from many local machines in which all the data are distributed. In the scenario when the query is more complex, the communication cost for sending it to all the local machines for processing can be very high. This paper aims to address this issue. Given a reference set of multiple and large-size patterns, we propose an approach to identifying its k nearest and farthest neighbors globally across all the local machines. By decomposing the reference patterns into a multi-resolution representation and using novel distance bound designs, our method guarantees the exact results in a communication-efficient manner. Analytical and empirical studies show that our method outperforms the state-of-the-art methods in saving significant bandwidth usage, especially for large numbers of machines and large-sized reference patterns.

This paper is based on the mass open and authentic data from the securities market with the statistical experimental method, to conduct test of MACD and MA expert systems for the securities trading software. The advantages and disadvantages of the two systems are compared by analysis which takes the winning rate, annual rate of return, and net profit margin as the management objectives on the basis of theoretical mathematical statistics. In addition to the annual transaction times, MA expert system is slightly better than MACD expert system.

Reverse tendency indexes of RSI and KDJ expert system are comparatively analyzed and the result that annual return rate and net profit rate of RSI expert system are superior to that of KDJ expert system for many times, in accordance with public and authentic mass data in securities market, with the use of statistical experiments, through testing RSI and KDJ expert system of securities trading software, with annual return rate, win rate, and net profit rate as management objectives and the theory of mathematical statistics as research basis.

This paper presents a low-power wireless ECG acquisition system-on-chip (SoC), including an RF front-end circuit, a power unit, an analog front-end circuit, and a digital circuitry. The proposed RF front-end circuit can provide the amplitude shift keying demodulation and distance to digital conversion to accurately receive the data from the reader. The received data will wake up the power unit to provide the required supply voltages of analog front-end (AFE) and digital circuitry. The AFE, including a pre-amplifier, an analog filter, a post-amplifier, and an analog-to-digital converter, is used for the ECG acquisition. Moreover, the EPC Class I Gen 2 UHF standard is employed in the digital circuitry for the handshaking of communication and the control of the system. The proposed SoC has been implemented in 0.18-μm standard CMOS process and the measured results reveal the communication is compatible to the RFID protocol. The average power consumption for the operating chip is 12 μW. Using a Sony PR44 battery to the supply power (605mAh@1.4V), the RFID tag SoC operates continuously for about 50,000 hours (>5 years), which is appropriate for wireless wearable ECG monitoring systems.