Pub Date : 2020-09-20DOI: 10.1109/ITSC45102.2020.9294470
Han Zhao, Huan Yang, Yu Wang, Danwei W. Wang, Rong Su
Traffic forecasting is of great importance to vehicle routing, traffic signal control and urban planning. However, traffic forecasting task is challenging due to several factors, such as complex spatial topological structure and dynamic changing of traffic status. Most existing methods have limited ability to capture both spatial and temporal dependence of traffic data. In this paper, we propose a novel end-to-end deep learning model, Attention based Graph Bi-LSTM networks (AGBN) to perform the traffic forecasting task. It uses graph convolutional network (GCN) to extract spatial features and bidirectional long short-term memory networks (Bi-LSTM) to capture the temporal dependence. The attention mechanism is used to select relevant features at all time steps. Experiments show that our model could extract both spatial and temporal dependence well and outperforms other baselines on real-world traffic datasets.
{"title":"Attention Based Graph Bi-LSTM Networks for Traffic Forecasting","authors":"Han Zhao, Huan Yang, Yu Wang, Danwei W. Wang, Rong Su","doi":"10.1109/ITSC45102.2020.9294470","DOIUrl":"https://doi.org/10.1109/ITSC45102.2020.9294470","url":null,"abstract":"Traffic forecasting is of great importance to vehicle routing, traffic signal control and urban planning. However, traffic forecasting task is challenging due to several factors, such as complex spatial topological structure and dynamic changing of traffic status. Most existing methods have limited ability to capture both spatial and temporal dependence of traffic data. In this paper, we propose a novel end-to-end deep learning model, Attention based Graph Bi-LSTM networks (AGBN) to perform the traffic forecasting task. It uses graph convolutional network (GCN) to extract spatial features and bidirectional long short-term memory networks (Bi-LSTM) to capture the temporal dependence. The attention mechanism is used to select relevant features at all time steps. Experiments show that our model could extract both spatial and temporal dependence well and outperforms other baselines on real-world traffic datasets.","PeriodicalId":394538,"journal":{"name":"2020 IEEE 23rd International Conference on Intelligent Transportation Systems (ITSC)","volume":"PP 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-09-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"126412235","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2020-09-20DOI: 10.1109/ITSC45102.2020.9294479
S. Anis, N. Sacco
Public transport of a region is extremely important for connecting the commuters from their origins to destinations. Public transport systems with large fleets cannot be guaranteed to perform efficiently, unless it is well connected and accessible to maximum possible population. In this regard, the localization of public transport stops (nodes) are highly important, since access to public transit systems is only possible through these nodes.In this framework, this paper focuses on the formulation of a general methodology for the evaluation of public transit nodes of a region based on transit system characteristics, spatial coverages and characteristics of zones using the concepts of connectivity and accessibility. Similarly, connectivity and accessibility index are calculated and enhanced based on distribution of public transport trips in zones and compared with each other to determine the critical nodes. To show the capability of the proposed approach, an application of this methodology in terms of a case study is analyzed in order to show the effects of trip distribution in the zones on the connectivity and accessibility index values during different time periods.
{"title":"Methodological Framework for the Evaluation of Critical Nodes in Public Transit Systems","authors":"S. Anis, N. Sacco","doi":"10.1109/ITSC45102.2020.9294479","DOIUrl":"https://doi.org/10.1109/ITSC45102.2020.9294479","url":null,"abstract":"Public transport of a region is extremely important for connecting the commuters from their origins to destinations. Public transport systems with large fleets cannot be guaranteed to perform efficiently, unless it is well connected and accessible to maximum possible population. In this regard, the localization of public transport stops (nodes) are highly important, since access to public transit systems is only possible through these nodes.In this framework, this paper focuses on the formulation of a general methodology for the evaluation of public transit nodes of a region based on transit system characteristics, spatial coverages and characteristics of zones using the concepts of connectivity and accessibility. Similarly, connectivity and accessibility index are calculated and enhanced based on distribution of public transport trips in zones and compared with each other to determine the critical nodes. To show the capability of the proposed approach, an application of this methodology in terms of a case study is analyzed in order to show the effects of trip distribution in the zones on the connectivity and accessibility index values during different time periods.","PeriodicalId":394538,"journal":{"name":"2020 IEEE 23rd International Conference on Intelligent Transportation Systems (ITSC)","volume":"19 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-09-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"126424192","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2020-09-20DOI: 10.1109/ITSC45102.2020.9294666
Chao Yan, C. Zheng, Chaoqian Gao, Wei Yu, Yuzhan Cai, Changjie Ma
Lane line is a very important element in HD maps, and map updating based on information can effectively reduce production cost. We use the images obtained by crowdsourcing for information mining. Most of these images are discontinuous and there are no internal or external parameters. However, lane detection algorithms are mostly applied to the vehicle, which are not suitable to detect road changed information. We propose a lane line perception network for information discovery, which directly takes the returned image as input and outputs the number of lane lines, as well as the color and type attributes of each lane. In contrast to previous works, we have solved the gradient explosion problem and specially optimized type segmentation. Finally, the proposed method is applied to mine information about lane changes.
{"title":"Lane Information Perception Network for HD Maps","authors":"Chao Yan, C. Zheng, Chaoqian Gao, Wei Yu, Yuzhan Cai, Changjie Ma","doi":"10.1109/ITSC45102.2020.9294666","DOIUrl":"https://doi.org/10.1109/ITSC45102.2020.9294666","url":null,"abstract":"Lane line is a very important element in HD maps, and map updating based on information can effectively reduce production cost. We use the images obtained by crowdsourcing for information mining. Most of these images are discontinuous and there are no internal or external parameters. However, lane detection algorithms are mostly applied to the vehicle, which are not suitable to detect road changed information. We propose a lane line perception network for information discovery, which directly takes the returned image as input and outputs the number of lane lines, as well as the color and type attributes of each lane. In contrast to previous works, we have solved the gradient explosion problem and specially optimized type segmentation. Finally, the proposed method is applied to mine information about lane changes.","PeriodicalId":394538,"journal":{"name":"2020 IEEE 23rd International Conference on Intelligent Transportation Systems (ITSC)","volume":"90 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-09-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"126439975","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2020-09-20DOI: 10.1109/ITSC45102.2020.9294471
Chia-Cheng Yen, D. Ghosal, Michael Zhang, C. Chuah
Reinforcement Learning (RL) is being rapidly adopted in many complex environments due to its ability to leverage neural networks to learn good strategies. In traffic signal control (TSC), existing work has focused on off-policy learning (Q-learning) with neural networks. There is limited study on on-policy learning (SARSA) with neural networks. In this work, we propose a deep dueling on-policy learning method (2DSARSA) for coordinated TSC for a network of intersections that maximizes the network throughput and minimizes the average end-to-end delay. To describe the states of the environment, we propose traffic flow maps (TFMs) that capture head-of-the-line (HOL) sojourn times for traffic lanes and HOL differences for adjacent intersections. We introduce a reward function defined by the power metric which is the ratio of the network throughput to the average end-to-end delay. The proposed reward function simultaneously maximizes the network throughput and minimizes the average end-to-end delay. We show that the proposed 2DSARSA architecture has a significantly better learning performance compared to other RL architectures including Deep Q-Network (DQN) and Deep SARSA (DSARSA).
{"title":"A Deep On-Policy Learning Agent for Traffic Signal Control of Multiple Intersections","authors":"Chia-Cheng Yen, D. Ghosal, Michael Zhang, C. Chuah","doi":"10.1109/ITSC45102.2020.9294471","DOIUrl":"https://doi.org/10.1109/ITSC45102.2020.9294471","url":null,"abstract":"Reinforcement Learning (RL) is being rapidly adopted in many complex environments due to its ability to leverage neural networks to learn good strategies. In traffic signal control (TSC), existing work has focused on off-policy learning (Q-learning) with neural networks. There is limited study on on-policy learning (SARSA) with neural networks. In this work, we propose a deep dueling on-policy learning method (2DSARSA) for coordinated TSC for a network of intersections that maximizes the network throughput and minimizes the average end-to-end delay. To describe the states of the environment, we propose traffic flow maps (TFMs) that capture head-of-the-line (HOL) sojourn times for traffic lanes and HOL differences for adjacent intersections. We introduce a reward function defined by the power metric which is the ratio of the network throughput to the average end-to-end delay. The proposed reward function simultaneously maximizes the network throughput and minimizes the average end-to-end delay. We show that the proposed 2DSARSA architecture has a significantly better learning performance compared to other RL architectures including Deep Q-Network (DQN) and Deep SARSA (DSARSA).","PeriodicalId":394538,"journal":{"name":"2020 IEEE 23rd International Conference on Intelligent Transportation Systems (ITSC)","volume":"7 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-09-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"125423714","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2020-09-20DOI: 10.1109/ITSC45102.2020.9294220
Hongsheng Qi
Today’s vehicles are capable of detecting environmental traffic participants, such as other vehicles, pedestrians, traffic lights etc, and communicating with each other or infrastructures. Typical on-board detectors include LiDAR, camera and so on. These vehicles which can make driving decisions based on the detected information without human intervention are named CAV (connected and autonomous vehicles). However, in a long period, the road traffic is mixed by traditional vehicles (human driven vehicles, or HVs) and CAV. The system can only “see” the near field vehicles around the CAVs by means of on-board detectors or VANET (vehicular ad hoc network). Far-field vehicles are either too far away or covered by near-field vehicles. In order to enhance the sensing capabilities of VANET or CAV, the manuscript propose a far-field vehicles sensing method, called F2-sensing. The method combines the deep learning and the car following logic. The rationale is that, as the vehicles react to downstream vehicles’ states variation, when the CAVs and the near field vehicles’ states are known, the downstream vehicles’ existence and its real-time location can be estimated. The proposed method is tested against real world dataset, which proves the usefulness of the method.
今天的车辆能够检测环境交通参与者，如其他车辆、行人、交通灯等，并与彼此或基础设施进行通信。典型的机载探测器包括激光雷达、摄像头等。这些基于检测到的信息而无需人为干预就能做出驾驶决策的车辆被称为CAV (connected and autonomous vehicles)。然而，在很长一段时间内，道路交通是由传统车辆(HVs)和自动驾驶汽车混合。该系统只能通过车载探测器或VANET(车载自组织网络)“看到”cav周围的近场车辆。远场车辆要么离得太远，要么被近场车辆覆盖。为了增强VANET或CAV的传感能力，本文提出了一种远场车辆传感方法，称为f2传感。该方法结合了深度学习和汽车跟随逻辑。其原理是，由于车辆对下游车辆状态变化的反应，当cav和近场车辆的状态已知时，可以估计下游车辆的存在及其实时位置。通过对实际数据集的测试，证明了该方法的有效性。
{"title":"Far-field sensing in partial VANET environment","authors":"Hongsheng Qi","doi":"10.1109/ITSC45102.2020.9294220","DOIUrl":"https://doi.org/10.1109/ITSC45102.2020.9294220","url":null,"abstract":"Today’s vehicles are capable of detecting environmental traffic participants, such as other vehicles, pedestrians, traffic lights etc, and communicating with each other or infrastructures. Typical on-board detectors include LiDAR, camera and so on. These vehicles which can make driving decisions based on the detected information without human intervention are named CAV (connected and autonomous vehicles). However, in a long period, the road traffic is mixed by traditional vehicles (human driven vehicles, or HVs) and CAV. The system can only “see” the near field vehicles around the CAVs by means of on-board detectors or VANET (vehicular ad hoc network). Far-field vehicles are either too far away or covered by near-field vehicles. In order to enhance the sensing capabilities of VANET or CAV, the manuscript propose a far-field vehicles sensing method, called F2-sensing. The method combines the deep learning and the car following logic. The rationale is that, as the vehicles react to downstream vehicles’ states variation, when the CAVs and the near field vehicles’ states are known, the downstream vehicles’ existence and its real-time location can be estimated. The proposed method is tested against real world dataset, which proves the usefulness of the method.","PeriodicalId":394538,"journal":{"name":"2020 IEEE 23rd International Conference on Intelligent Transportation Systems (ITSC)","volume":"25 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-09-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"125576272","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2020-09-20DOI: 10.1109/ITSC45102.2020.9294431
Yuying Shen, Hai-ying Li, Xin-yue Xu
Travel time variability is a feature of transport systems, which adds additional costs and uncertainty to travelers. Hence, it should be given greater emphasis on appraisal studies. This paper analyzes the travel time volatility of Origin-Destination (OD) pairs with different characteristics in different time of day, and develop a two-level approach to evaluate travel time variability in metros. The average travel time of network is calculated to assess the efficiency of metro operation from the network-wide level and the coefficient of variation is used to identify the abnormal OD pairs with travel time variability from OD level. Based on the tap-in and tap-out data collected by automated fare collection (AFC) facilities as well as existing unexpected events within networks, an analysis of travel time variability is undertaken in Beijing metro to validate the proposed method.
{"title":"Evaluation of travel time variability in metros based on AFC data*","authors":"Yuying Shen, Hai-ying Li, Xin-yue Xu","doi":"10.1109/ITSC45102.2020.9294431","DOIUrl":"https://doi.org/10.1109/ITSC45102.2020.9294431","url":null,"abstract":"Travel time variability is a feature of transport systems, which adds additional costs and uncertainty to travelers. Hence, it should be given greater emphasis on appraisal studies. This paper analyzes the travel time volatility of Origin-Destination (OD) pairs with different characteristics in different time of day, and develop a two-level approach to evaluate travel time variability in metros. The average travel time of network is calculated to assess the efficiency of metro operation from the network-wide level and the coefficient of variation is used to identify the abnormal OD pairs with travel time variability from OD level. Based on the tap-in and tap-out data collected by automated fare collection (AFC) facilities as well as existing unexpected events within networks, an analysis of travel time variability is undertaken in Beijing metro to validate the proposed method.","PeriodicalId":394538,"journal":{"name":"2020 IEEE 23rd International Conference on Intelligent Transportation Systems (ITSC)","volume":"17 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-09-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"127949672","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2020-09-20DOI: 10.1109/ITSC45102.2020.9294737
Carlos Gómez Huélamo, Javier del Egido, L. Bergasa, R. Barea, M. Ocaña, J. F. Arango, Rodrigo Gutiérrez-Moreno
This paper presents a Real-Time Bird’s Eye View Multi Object Tracking (MOT) system pipeline for an Autonomous Electric car, based on Fast Encoders for object detection and a combination of Hungarian algorithm and Bird’s Eye View (BEV) Kalman Filter, respectively used for data association and state estimation. The system is able to analyze 360 degrees around the ego-vehicle as well as estimate the future trajectories of the environment objects, being the essential input for other layers of a self-driving architecture, such as the control or decision-making. First, our system pipeline is described, merging the concepts of online and real-time DATMO (Deteccion and Tracking of Multiple Objects), ROS (Robot Operating System) and Docker to enhance the integration of the proposed MOT system in fully-autonomous driving architectures. Second, the system pipeline is validated using the recently proposed KITTI-3DMOT evaluation tool that demonstrates the full strength of 3D localization and tracking of a MOT system. Finally, a comparison of our proposal with other state-of-the-art approaches is carried out in terms of performance by using the mainstream metrics used on MOT benchmarks and the recently proposed integral MOT metrics, evaluating the performance of the tracking system over all detection thresholds.
{"title":"Real-Time Bird’s Eye View Multi-Object Tracking system based on Fast Encoders for Object Detection","authors":"Carlos Gómez Huélamo, Javier del Egido, L. Bergasa, R. Barea, M. Ocaña, J. F. Arango, Rodrigo Gutiérrez-Moreno","doi":"10.1109/ITSC45102.2020.9294737","DOIUrl":"https://doi.org/10.1109/ITSC45102.2020.9294737","url":null,"abstract":"This paper presents a Real-Time Bird’s Eye View Multi Object Tracking (MOT) system pipeline for an Autonomous Electric car, based on Fast Encoders for object detection and a combination of Hungarian algorithm and Bird’s Eye View (BEV) Kalman Filter, respectively used for data association and state estimation. The system is able to analyze 360 degrees around the ego-vehicle as well as estimate the future trajectories of the environment objects, being the essential input for other layers of a self-driving architecture, such as the control or decision-making. First, our system pipeline is described, merging the concepts of online and real-time DATMO (Deteccion and Tracking of Multiple Objects), ROS (Robot Operating System) and Docker to enhance the integration of the proposed MOT system in fully-autonomous driving architectures. Second, the system pipeline is validated using the recently proposed KITTI-3DMOT evaluation tool that demonstrates the full strength of 3D localization and tracking of a MOT system. Finally, a comparison of our proposal with other state-of-the-art approaches is carried out in terms of performance by using the mainstream metrics used on MOT benchmarks and the recently proposed integral MOT metrics, evaluating the performance of the tracking system over all detection thresholds.","PeriodicalId":394538,"journal":{"name":"2020 IEEE 23rd International Conference on Intelligent Transportation Systems (ITSC)","volume":"25 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-09-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"132229413","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2020-09-20DOI: 10.1109/ITSC45102.2020.9294482
Cong Fei, Xiangkun He, Sadahiro Kawahara, S. Nakano, Xuewu Ji
Trajectory prediction is a crucial task required for autonomous driving. The highly interactions and uncertainties in real-world traffic scenarios make it a challenge to generate trajectories that are accurate, reasonable and covering diverse modality as much as possible. This paper propose a conditional Wasserstein auto-encoder trajectory prediction model (TrajCWAE) that combines the representation learning and variational inference to generate predictions with multi-modal nature. TrajCWAE model leverages a context embedder to learn the intentions among vehicles and imposes Gaussian mixture model to reconstruct the prior and posterior distributions. Wasserstein generative adversarial framework is then used to match the aggregated posterior distribution with prior distribution. Furthermore, kinematic constraints are considered to make the prediction physically feasible and socially acceptable. Experiments on two scenarios demonstrate that the proposed model outperforms state-of-the-art methods, achieving better accuracy, diversity and coverage.
{"title":"Conditional Wasserstein Auto-Encoder for Interactive Vehicle Trajectory Prediction","authors":"Cong Fei, Xiangkun He, Sadahiro Kawahara, S. Nakano, Xuewu Ji","doi":"10.1109/ITSC45102.2020.9294482","DOIUrl":"https://doi.org/10.1109/ITSC45102.2020.9294482","url":null,"abstract":"Trajectory prediction is a crucial task required for autonomous driving. The highly interactions and uncertainties in real-world traffic scenarios make it a challenge to generate trajectories that are accurate, reasonable and covering diverse modality as much as possible. This paper propose a conditional Wasserstein auto-encoder trajectory prediction model (TrajCWAE) that combines the representation learning and variational inference to generate predictions with multi-modal nature. TrajCWAE model leverages a context embedder to learn the intentions among vehicles and imposes Gaussian mixture model to reconstruct the prior and posterior distributions. Wasserstein generative adversarial framework is then used to match the aggregated posterior distribution with prior distribution. Furthermore, kinematic constraints are considered to make the prediction physically feasible and socially acceptable. Experiments on two scenarios demonstrate that the proposed model outperforms state-of-the-art methods, achieving better accuracy, diversity and coverage.","PeriodicalId":394538,"journal":{"name":"2020 IEEE 23rd International Conference on Intelligent Transportation Systems (ITSC)","volume":"33 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-09-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"132312796","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2020-09-20DOI: 10.1109/ITSC45102.2020.9294244
Yuta Kataoka, Hao Yang, Shalini Keshavamurthy, Ippei Nishitani, K. Oguchi
Reinforcement learning is one of the methods that has been used to realize optimal driving. Most studies have focused on evaluating learning performance of a fraction of vehicles controlled by reinforcement learning. It is unclear how these controlled vehicles influence other vehicles. We conducted several experiments examining the impact of multiple vehicles controlled by reinforcement learning on traffic flow. The simulations were performed on a three-lane freeway with lane regulation at the end of one of the lanes. The controlled vehicles were trained to drive as fast as possible and run non-cooperatively. We found out that controlled vehicles could run faster than human-driven vehicles. Moreover, we anticipated that if multiple vehicles were run selfishly, it would adversely affect traffic flow. Contrary to expectations, the experimental results showed that even if numerous controlled vehicles drive selfishly, the negative impact on overall traffic would be small.
{"title":"Traffic Impact Analysis of a Deep Reinforcement Learning-based Multi-lane Freeway Vehicle Control","authors":"Yuta Kataoka, Hao Yang, Shalini Keshavamurthy, Ippei Nishitani, K. Oguchi","doi":"10.1109/ITSC45102.2020.9294244","DOIUrl":"https://doi.org/10.1109/ITSC45102.2020.9294244","url":null,"abstract":"Reinforcement learning is one of the methods that has been used to realize optimal driving. Most studies have focused on evaluating learning performance of a fraction of vehicles controlled by reinforcement learning. It is unclear how these controlled vehicles influence other vehicles. We conducted several experiments examining the impact of multiple vehicles controlled by reinforcement learning on traffic flow. The simulations were performed on a three-lane freeway with lane regulation at the end of one of the lanes. The controlled vehicles were trained to drive as fast as possible and run non-cooperatively. We found out that controlled vehicles could run faster than human-driven vehicles. Moreover, we anticipated that if multiple vehicles were run selfishly, it would adversely affect traffic flow. Contrary to expectations, the experimental results showed that even if numerous controlled vehicles drive selfishly, the negative impact on overall traffic would be small.","PeriodicalId":394538,"journal":{"name":"2020 IEEE 23rd International Conference on Intelligent Transportation Systems (ITSC)","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-09-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"130109713","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2020-09-20DOI: 10.1109/ITSC45102.2020.9294385
Sheng-Rong Zhao, S. Lin, Yunlong Li, Jungang Xu, Yibing Wang
Predicting urban traffic flow is significant to intelligent transportation systems. Urban traffic flow data is a type of time-series data, which collects the traffic flow of a road section or area. So in this paper, we treat urban traffic flow prediction as a time series problem. The traditional method to tackle traffic flow prediction is difficult, because of the complex influence factors and nonlinear dependencies. Recently, LSTM based network has been widely used to model long-term series, but the memory of LSTM is typically too small and is not enough to accurately remember facts from the past. In this paper, we using memory time-series network with additional memory mechanisms to address urban traffic prediction problems. Historical data were divided into longterm and short-term two parts, long-term historical data models the overall trend and short-term historical data takes into account recent changes. The experiment results on two urban traffic flow datasets demonstrate the model is effective and outperforms baselines.
{"title":"Urban Traffic Flow Forecasting Based on Memory Time-Series Network","authors":"Sheng-Rong Zhao, S. Lin, Yunlong Li, Jungang Xu, Yibing Wang","doi":"10.1109/ITSC45102.2020.9294385","DOIUrl":"https://doi.org/10.1109/ITSC45102.2020.9294385","url":null,"abstract":"Predicting urban traffic flow is significant to intelligent transportation systems. Urban traffic flow data is a type of time-series data, which collects the traffic flow of a road section or area. So in this paper, we treat urban traffic flow prediction as a time series problem. The traditional method to tackle traffic flow prediction is difficult, because of the complex influence factors and nonlinear dependencies. Recently, LSTM based network has been widely used to model long-term series, but the memory of LSTM is typically too small and is not enough to accurately remember facts from the past. In this paper, we using memory time-series network with additional memory mechanisms to address urban traffic prediction problems. Historical data were divided into longterm and short-term two parts, long-term historical data models the overall trend and short-term historical data takes into account recent changes. The experiment results on two urban traffic flow datasets demonstrate the model is effective and outperforms baselines.","PeriodicalId":394538,"journal":{"name":"2020 IEEE 23rd International Conference on Intelligent Transportation Systems (ITSC)","volume":"17 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-09-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"130449987","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
京公网安备 11010802042870号
京ICP备2023020795号-1
扫码关注我们
求助内容：
应助结果提醒方式：