Traffic sign recognition based on multi-scale convolutional neural network
Accurate traffic sign recognition is the key problem to realize automatic driving. In order to overcome the impact of lighting, geographical location, detection methods and other factors on traffic sign recognition in the actual scene, and improve the accuracy of sign recognition. In this paper, a traffic sign recognition based on multi-scale convolutional neural network is provided. Firstly , a new multi-scale convolutional neural network is present and verified. Then, traffic sign recognition is carried out using the new multi-scale convolutional neural network. The experimental results show that the average accuracy of traffic sign recognition reaches 89.5% The classification accuracy reaches 90.2%, the prohibition classification accuracy reaches 91.25%, the direction classification accuracy reaches 82.12%, and the indication classification accuracy reaches 90.68%.
Application of edge computing in intelligent transportation system
The problem of traffic congestion is becoming more and more serious. Although the intelligent transportation system relying on traditional cloud computing can alleviate the traffic pressure to a certain extent, it can no longer meet the requirements of transmission bandwidth and delay for new on-board applications such as assisted driving and automatic driving. In order to realize the real-time processing of massive data, ensure public information and traffic safety, and improve the operation efficiency of transportation system, edge computing is applied to intelligent transportation. Firstly, the development of intelligent transportation is described as a whole, and the overall architecture of Intelligent Transportation Based on edge computing is proposed, which makes full use of the characteristics of physical proximity, high bandwidth, low delay and location cognition of edge computing to solve the problems of information transmission delay, untimely data processing and large transmission load of the current transportation system. Then, based on wireless transmission, information perception, computing unloading and collaborative processing, this paper expounds the key technologies to be solved in the application of edge computing in intelligent transportation. Finally, it points out the future opportunities and challenges faced by the application of edge computing in intelligent transportation.
Vehicle route planning based on traffic big data
In the future The application and development of traffic big data has brought opportunities and challenges to modern vehicle routing planning. Therefore, understanding the research status and characteristics of traffic big data concept, road network matching, path planning algorithm and traffic information prediction is particularly important to clarify the research direction and development trend of path planning in the future. Firstly, the concept of traffic big data and the preprocessing method of trajectory data are introduced, and various matching algorithms and their advantages and disadvantages in road network matching at home and abroad are summarized; Then, the common path planning algorithms are described, including traditional classical algorithms and popular intelligent algorithms; Then it briefly summarizes the research methods and various prediction models of traffic information prediction; Finally, the existing problems of vehicle routing planning at the present stage are pointed out, and the future research direction is prospected.
Predicting Freeway Travel Time Using Multiple-Source Heterogeneous Data Integration
Freeway travel time is influenced by many factors including traffic volume, adverse weather, accidents, traffic control, and so on. We employ the multiple source data-mining method to analyze freeway travel time. We collected toll data, weather data, traffic accident disposal logs, and other historical data from Freeway G5513 in Hunan Province, China. Using the Support Vector Machine (SVM), we proposed the travel time predicting model founded on these databases. The new SVM model can simulate the nonlinear relationship between travel time and those factors. In order to improve the precision of the SVM model, we applied the Artificial Fish Swarm algorithm to optimize the SVM model parameters, which include the kernel parameter σ, non-sensitive loss function parameter ε, and penalty parameter C. We compared the new optimized SVM model with the Back Propagation (BP) neural network and a common SVM model, using the historical data collected from freeway G5513. The results show that the accuracy of the optimized SVM model is 17.27% and 16.44% higher than those of the BP neural network model and the common SVM model, respectively.
Achieving High Accuracy Localization with RIS: Challenges and Opportunities
Location-based services (LBS) are indispensable for future smart cities, but the dense urban built environment is a typical multipath propagation scenario in which it is challenging to achieve accurate real-time localization. The emergence of Reconfigurable Intelligent Surfaces (RIS) offers new hope for solving this problem. RIS is able to achieve anomalous propagation of electromagnetic waves using metamaterial-based planar or conformal large surfaces，in other words, by designing controllable wireless channels, Smart Radio Environments(SRE) can be purposefully constructed, hence, RIS becoming a strong candidate for key technologies in the physical layer of 6G. Current work around RIS mainly focuses on maximizing energy efficiency, low-complexity channel estimation, beamforming optimization, and physical layer security, but the application of RIS to complex environments to enhance localization accuracy has not received the attention it deserves. In this report, we provide a brief review of the RIS-enhanced simultaneous communication localization system, analyzing the key technologies and implementation challenges involved, and, in particular, we highlight the ability of RIS to significantly improve the localization capabilities of randomly deployed Internet of Things (IoT).