The present paper aims to describe the conceptual idea to use cars as sensors to measure and acquire data related road environment. The parameters are collected using only standard equipment commonly installed and ope...The present paper aims to describe the conceptual idea to use cars as sensors to measure and acquire data related road environment. The parameters are collected using only standard equipment commonly installed and operative on commercial cars. Real sensors and car sub-systems (e.g. thermometers, accelerometers, ABS, ESP, and GPS) together with other “implicit” sensors (e.g. fog lights, windscreen wipers) acquire and contain information. They are shared inside an in-vehicle communication network using mainly the standard CAN bus and can be collected by a simple central node. This node can also be available on the market without too expensive costs thanks to some companies which business is devoted to car fleet monitoring. All the collected data are then geolocalized using a standard GPS receiver and sent to a remote elaboration unit, exploiting mobile network technologies such as GPRS or UMTS. A large number of cars, connected together in a diffuse Wireless Sensor Network, allow the elaboration unit to realize some info-layers put at the disposal of a car driver. Traffic, state of the road and other information about the weather can be received by car drivers using an ad hoc developed mobile application for smartphone which can give punctual information related to a specific route, previously set on the mobile phone navigator. The description of some experimental activities is presented, some technical points will be addressed and some examples of applications of the network of cars “as sensors” will be given.展开更多
车辆型号识别在智能交通系统、涉车刑侦案件侦破等方面具有十分重要的应用前景.针对车辆型号种类繁多、部分型号区分度小等带来的车辆型号精细分类困难的问题,采用车辆正脸图像为数据源,提出一种多分支多维度特征融合的卷积神经网络模型...车辆型号识别在智能交通系统、涉车刑侦案件侦破等方面具有十分重要的应用前景.针对车辆型号种类繁多、部分型号区分度小等带来的车辆型号精细分类困难的问题,采用车辆正脸图像为数据源,提出一种多分支多维度特征融合的卷积神经网络模型Fg-CarNet (Convolutional neural networks for car fine-grained classification, Fg-CarNet).该模型根据车正脸图像特征分布特点,将其分为上下两部分并行进行特征提取,并对网络中间层产生的特征进行两个维度的融合,以提取有区分度的特征,提高特征表达能力,通过使用小卷积核以及全局均值池化,使在网络分类准确度提高的同时降低了网络模型参数大小.在CompCars数据集上进行验证,实验结果表明, Fg-CarNet提取的车辆特征在保证网络模型参数最小的同时,车辆型号识别率达到最高,实现了最好的分类效果.展开更多
We introduce an almost-automatic technique for generating 3D car styling surface models based on a single side-view image. Our approach combines the prior knowledge of car styling and deformable curve network model to...We introduce an almost-automatic technique for generating 3D car styling surface models based on a single side-view image. Our approach combines the prior knowledge of car styling and deformable curve network model to obtain an automatic modeling process. Firstly, we define the consistent parameterized curve template for 2D and 3D case respectively by analyzing the characteristic lines for car styling. Then, a semi-automatic extraction from a side-view car image is adopted. Thirdly, statistic morphable model of 3D curve network is used to get the initial solution with sparse point constraints. With only a few post-processing operations, the optimized curve network models for creating surfaces are obtained. Finally, the styling surfaces are automatically generated using template-based parametric surface modeling method. More than 50 3D curve network models are constructed as the morphable database. We show that this intelligent modeling tool simplifies the exhausted modeling task, and also demonstrate meaningful results of our approach.展开更多
The decision-making under complex urban environment become one of the key issues that restricts the rapid development of the autonomous vehicles. The difficulty in making timely and accurate decisions like human being...The decision-making under complex urban environment become one of the key issues that restricts the rapid development of the autonomous vehicles. The difficulty in making timely and accurate decisions like human beings under highly dynamic traffic environment is a major challenge for autonomous driving. Car-following has been regarded as the simplest but essential driving behavior among driving tasks and has received extensive attention from researchers around the world. This work addresses this problem and proposes a novel method RSAN(rough-set artificial neural network) to learn the decisions from excellent human drivers. A virtual urban traffic environment was built by Pre Scan and driving simulation was conducted to obtain a broad set of relevant data such as experienced drivers' behavior data and surrounding vehicles' motion data. Then, rough set was used to preprocess these data to extract the key influential factors on decision and reduce the impact of uncertain data and noise data. And the car-following decision was learned by neural network in which key factor was the input and acceleration was the output. The result shows the better convergence speed and the better decision accuracy of RSAN than ANN. Findings of this work contributes to the empirical understanding of driver's decision-making process and it provides a theoretical basis for the study of car-following decision-making under complex and dynamic environment.展开更多
This paper discusses the dynamic behavior and its predictions for a simulated traffic flow based on the nonlinear response of a vehicle to the leading car’s movement in a single lane.Traffic chaos is a promising fiel...This paper discusses the dynamic behavior and its predictions for a simulated traffic flow based on the nonlinear response of a vehicle to the leading car’s movement in a single lane.Traffic chaos is a promising field,and chaos theory has been applied to identify and predict its chaotic movement.A simulated traffic flow is generated using a car-following model(GM model),and the distance between two cars is investigated for its dynamic properties.A positive Lyapunov exponent confirms the existence of chaotic behavior in the GM model.A new algorithm using a RBF NN (radial basis function neural network) is proposed to predict this traffic chaos.The experiment shows that the chaotic degree and predictable degree are determined by the first Lyapunov exponent.The algorithm proposed in this paper can be generalized to recognize and predict the chaos of short-time traffic flow {series.}展开更多
After introducing the principle of float car data (FCD), this paper gives the primary flow of pre-handing and map- matching of the FCD. After analyzing the percentage of coverage of FCD on the road network, large quan...After introducing the principle of float car data (FCD), this paper gives the primary flow of pre-handing and map- matching of the FCD. After analyzing the percentage of coverage of FCD on the road network, large quantity of heritage database of routing status is used to estimate the routing velocity when lack of FCD on parts road segments. Multi liner regression model is then put forwarded by considering the spatial correlativity among the road network, and some model parameters are deduced when time series is classified in day and week. Besides, error of velocity probability and error of status probability are achieved based on the result from field testing while the feasibility and reliability of the velocity estimation model is obtained as well. Finally, as a case study in Shanghai center area, the whole routing velocity in the road network is estimated and published in real time.展开更多
文摘The present paper aims to describe the conceptual idea to use cars as sensors to measure and acquire data related road environment. The parameters are collected using only standard equipment commonly installed and operative on commercial cars. Real sensors and car sub-systems (e.g. thermometers, accelerometers, ABS, ESP, and GPS) together with other “implicit” sensors (e.g. fog lights, windscreen wipers) acquire and contain information. They are shared inside an in-vehicle communication network using mainly the standard CAN bus and can be collected by a simple central node. This node can also be available on the market without too expensive costs thanks to some companies which business is devoted to car fleet monitoring. All the collected data are then geolocalized using a standard GPS receiver and sent to a remote elaboration unit, exploiting mobile network technologies such as GPRS or UMTS. A large number of cars, connected together in a diffuse Wireless Sensor Network, allow the elaboration unit to realize some info-layers put at the disposal of a car driver. Traffic, state of the road and other information about the weather can be received by car drivers using an ad hoc developed mobile application for smartphone which can give punctual information related to a specific route, previously set on the mobile phone navigator. The description of some experimental activities is presented, some technical points will be addressed and some examples of applications of the network of cars “as sensors” will be given.
文摘车辆型号识别在智能交通系统、涉车刑侦案件侦破等方面具有十分重要的应用前景.针对车辆型号种类繁多、部分型号区分度小等带来的车辆型号精细分类困难的问题,采用车辆正脸图像为数据源,提出一种多分支多维度特征融合的卷积神经网络模型Fg-CarNet (Convolutional neural networks for car fine-grained classification, Fg-CarNet).该模型根据车正脸图像特征分布特点,将其分为上下两部分并行进行特征提取,并对网络中间层产生的特征进行两个维度的融合,以提取有区分度的特征,提高特征表达能力,通过使用小卷积核以及全局均值池化,使在网络分类准确度提高的同时降低了网络模型参数大小.在CompCars数据集上进行验证,实验结果表明, Fg-CarNet提取的车辆特征在保证网络模型参数最小的同时,车辆型号识别率达到最高,实现了最好的分类效果.
基金Supported by National Natural Science Foundation of China(Nos.11472073,61173102 and 61370143)
文摘We introduce an almost-automatic technique for generating 3D car styling surface models based on a single side-view image. Our approach combines the prior knowledge of car styling and deformable curve network model to obtain an automatic modeling process. Firstly, we define the consistent parameterized curve template for 2D and 3D case respectively by analyzing the characteristic lines for car styling. Then, a semi-automatic extraction from a side-view car image is adopted. Thirdly, statistic morphable model of 3D curve network is used to get the initial solution with sparse point constraints. With only a few post-processing operations, the optimized curve network models for creating surfaces are obtained. Finally, the styling surfaces are automatically generated using template-based parametric surface modeling method. More than 50 3D curve network models are constructed as the morphable database. We show that this intelligent modeling tool simplifies the exhausted modeling task, and also demonstrate meaningful results of our approach.
基金Project(9142020013)support by the National Natural Science Foundation of China
文摘The decision-making under complex urban environment become one of the key issues that restricts the rapid development of the autonomous vehicles. The difficulty in making timely and accurate decisions like human beings under highly dynamic traffic environment is a major challenge for autonomous driving. Car-following has been regarded as the simplest but essential driving behavior among driving tasks and has received extensive attention from researchers around the world. This work addresses this problem and proposes a novel method RSAN(rough-set artificial neural network) to learn the decisions from excellent human drivers. A virtual urban traffic environment was built by Pre Scan and driving simulation was conducted to obtain a broad set of relevant data such as experienced drivers' behavior data and surrounding vehicles' motion data. Then, rough set was used to preprocess these data to extract the key influential factors on decision and reduce the impact of uncertain data and noise data. And the car-following decision was learned by neural network in which key factor was the input and acceleration was the output. The result shows the better convergence speed and the better decision accuracy of RSAN than ANN. Findings of this work contributes to the empirical understanding of driver's decision-making process and it provides a theoretical basis for the study of car-following decision-making under complex and dynamic environment.
文摘This paper discusses the dynamic behavior and its predictions for a simulated traffic flow based on the nonlinear response of a vehicle to the leading car’s movement in a single lane.Traffic chaos is a promising field,and chaos theory has been applied to identify and predict its chaotic movement.A simulated traffic flow is generated using a car-following model(GM model),and the distance between two cars is investigated for its dynamic properties.A positive Lyapunov exponent confirms the existence of chaotic behavior in the GM model.A new algorithm using a RBF NN (radial basis function neural network) is proposed to predict this traffic chaos.The experiment shows that the chaotic degree and predictable degree are determined by the first Lyapunov exponent.The algorithm proposed in this paper can be generalized to recognize and predict the chaos of short-time traffic flow {series.}
文摘After introducing the principle of float car data (FCD), this paper gives the primary flow of pre-handing and map- matching of the FCD. After analyzing the percentage of coverage of FCD on the road network, large quantity of heritage database of routing status is used to estimate the routing velocity when lack of FCD on parts road segments. Multi liner regression model is then put forwarded by considering the spatial correlativity among the road network, and some model parameters are deduced when time series is classified in day and week. Besides, error of velocity probability and error of status probability are achieved based on the result from field testing while the feasibility and reliability of the velocity estimation model is obtained as well. Finally, as a case study in Shanghai center area, the whole routing velocity in the road network is estimated and published in real time.