Article
Multidisciplinary Sciences
Dongjiang Liu, Leixiao Li
Summary: Intelligent traffic light control (ITLC) algorithms are efficient for relieving traffic congestion. Recent research focuses on improving reinforcement learning and coordination methods for decentralized multi-agent traffic light control. However, communication details need to be improved for effective coordination. Additionally, the traditional reward calculation method should consider both waiting time and queue length. Therefore, a new ITLC algorithm is proposed, addressing these problems by improving communication efficiency and introducing a new reward calculation method.
SCIENTIFIC REPORTS
(2023)
Article
Computer Science, Information Systems
Ruijie Zhu, Lulu Li, Shuning Wu, Pei Lv, Yafei Li, Mingliang Xu
Summary: Intelligent traffic light control aims to alleviate traffic congestion. This paper proposes a multi-agent broad reinforcement learning (MABRL) algorithm for traffic light control, which utilizes a broad network and a dynamic interaction mechanism to improve training efficiency and alleviate traffic congestion.
INFORMATION SCIENCES
(2023)
Article
Computer Science, Information Systems
Faizan Rasheed, Kok-Lim Alvin Yau, Rafidah Md Noor, Yung-Wey Chong
Summary: This paper investigates the use of multi-agent deep Q-network (MADQN) to address the curse of dimensionality issue in traditional multi-agent reinforcement learning (MARL), and conducts case studies on real traffic networks and grid traffic networks. The results show that the MADQN scheme has a significant effect on traffic signal control.
CMC-COMPUTERS MATERIALS & CONTINUA
(2022)
Article
Transportation Science & Technology
Mobin Yazdani, Majid Sarvi, Saeed Asadi Bagloee, Neema Nassir, Jeff Price, Hossein Parineh
Summary: This study proposes a novel deep RL-based adaptive traffic signal model that aims to minimize total user delays by allocating equitable green time for vehicles and pedestrians. The proposed intelligent vehicle pedestrian light (IVPL) method can handle both pedestrian and vehicle flows, even in the presence of jaywalking. Experimental results using a microsimulation model of an intersection in Melbourne demonstrate the superiority of the proposed model in terms of optimal solution quality and total user delay minimization.
TRANSPORTATION RESEARCH PART C-EMERGING TECHNOLOGIES
(2023)
Article
Engineering, Chemical
Bin Wang, Zhengkun He, Jinfang Sheng, Yu Chen
Summary: This paper proposes a traffic light timing optimization method called EP-D3QN based on double dueling deep Q-network, MaxPressure, and Self-organizing traffic lights (SOTL). The method controls traffic flows by dynamically adjusting the duration of traffic lights in a cycle, leading to significant reductions in waiting and travel times for vehicles, and improving the efficiency of intersections.
Article
Automation & Control Systems
Ming Yang, Yiming Wang, Yang Yu, Mingliang Zhou, U. Leong Hou
Summary: This article discusses the method of learning traffic light configuration in a mixed policy environment. The authors propose an executor-guide dual network and an improved centralized training and decentralized execution framework, and the experimental results demonstrate the superiority of this method.
IEEE TRANSACTIONS ON INDUSTRIAL INFORMATICS
(2023)
Article
Chemistry, Multidisciplinary
Qiming Zheng, Hongfeng Xu, Jingyun Chen, Dong Zhang, Kun Zhang, Guolei Tang
Summary: This study focuses on the real-time isolated signal control (RISC) problem at an intersection and improves a prevailing reinforcement learning (RL) method to solve it. By considering traffic engineering considerations and applying qualitative applicability analysis, the researchers propose a new RL algorithm based on deep Q-network (DDQN) and temporal-difference algorithm TD(Dyn) to address the problem. Experimental results demonstrate that the proposed method, termed D3ynQN, effectively reduces average vehicle delay compared to traditional fully-actuated control techniques.
APPLIED SCIENCES-BASEL
(2022)
Article
Computer Science, Artificial Intelligence
Min Wang, Libing Wu, Man Li, Dan Wu, Xiaochuan Shi, Chao Ma
Summary: Traffic signal control is crucial for urban transportation systems and public travel, but it becomes challenging due to the presence of spatial-temporal correlations and dynamic changes. If these issues are not resolved, it will lead to increased traffic pressure and wasted time.
KNOWLEDGE-BASED SYSTEMS
(2022)
Article
Chemistry, Multidisciplinary
Yang Shi, Zhenbo Wang, Tim J. LaClair, Chieh (Ross) Wang, Yunli Shao, Jinghui Yuan
Summary: The advent of connected vehicle technology brings new possibilities for revolutionizing future transportation systems. This paper proposes a novel data-driven traffic signal control method that combines deep learning and reinforcement learning techniques. By incorporating a compressed representation of the traffic states, the proposed method overcomes limitations in defining the action space, offering more practical and flexible signal phases. Simulation results demonstrate the convergence and robust performance of the proposed method compared to existing benchmark methods.
APPLIED SCIENCES-BASEL
(2023)
Article
Engineering, Civil
Soheil Mohamad Alizadeh Shabestary, Baher Abdulhai
Summary: This study proposes an adaptive traffic signal controller that utilizes deep learning techniques to process high-dimensional sensory inputs, aiming to optimize traffic flow and reduce delay time.
IEEE TRANSACTIONS ON INTELLIGENT TRANSPORTATION SYSTEMS
(2022)
Article
Engineering, Electrical & Electronic
Zixin Wang, Hanyu Zhu, Mingcheng He, Yong Zhou, Xiliang Luo, Ning Zhang
Summary: By using a combination of generative adversarial network (GAN) and multi-agent deep reinforcement learning (DRL), we propose a decentralized ATSC framework that enables autonomous control and collaboration of traffic signals by exchanging traffic statistics. This framework is scalable to large-scale traffic networks and robust to traffic flow variations.
IEEE TRANSACTIONS ON VEHICULAR TECHNOLOGY
(2022)
Article
Computer Science, Information Systems
Naoki Kodama, Taku Harada, Kazuteru Miyazaki
Summary: This article discusses the research progress on the use of deep reinforcement learning for traffic signal control. The authors propose a traffic light control system based on the dual targeting algorithm and validate its advantage over conventional methods through experiments.
Article
Computer Science, Information Systems
Xiangxue Zhao, Dominic Flocco, Shapour Azarm, Balakumar Balachandran
Summary: Reinforcement Learning (RL) is a popular approach for optimizing traffic signal control policy to alleviate congestion in a road network. This study proposes a new RL-based technique that co-optimizes the design of vehicular flow directions and control policy for traffic signals. By iteratively eliminating poor performing flow directions and generating new ones, the proposed technique aims to achieve better traffic performance and convergence to maximum expected performance.
Article
Chemistry, Analytical
Nagaiah Mohanan Balamurugan, Malaiyalathan Adimoolam, Mohammed H. Alsharif, Peerapong Uthansakul
Summary: Network traffic pattern identification and analysis are crucial for meeting different needs. This study introduces a new method based on an enhanced deep reinforcement learning algorithm to improve the accuracy and precision of network traffic analysis and prediction. Experimental results show that the proposed algorithm outperforms traditional convolutional neural network algorithms in terms of false positive and negative rates.
Article
Computer Science, Information Systems
Ruijie Zhu, Shuning Wu, Lulu Li, Ping Lv, Mingliang Xu
Summary: This article introduces a context-aware multiagent control method based on broad reinforcement learning for traffic light control. In comparison with previous methods, it takes into consideration pedestrian waiting states and adjacent agent states, effectively alleviating traffic congestion.
IEEE INTERNET OF THINGS JOURNAL
(2022)
Article
Computer Science, Information Systems
Jie Tian, Xiaoyuan Liang, Guiling Wang
Article
Computer Science, Information Systems
Xiaoyuan Liang, Tan Yan, Joyoung Lee, Guiling Wang
IEEE INTERNET OF THINGS JOURNAL
(2018)
Article
Computer Science, Information Systems
Jie Tian, Xiaoyuan Liang, Tan Yan, Mahesh Kumar Somashekar, Guiling Wang, Cesar Bandera
Article
Engineering, Civil
Xiaoyuan Liang, Yuchuan Zhang, Guiling Wang, Songhua Xu
IEEE TRANSACTIONS ON INTELLIGENT TRANSPORTATION SYSTEMS
(2020)
Proceedings Paper
Computer Science, Information Systems
Xiaoyuan Liang, Guiling Wang, Zhu Han
IEEE 2018 INTERNATIONAL CONGRESS ON CYBERMATICS / 2018 IEEE CONFERENCES ON INTERNET OF THINGS, GREEN COMPUTING AND COMMUNICATIONS, CYBER, PHYSICAL AND SOCIAL COMPUTING, SMART DATA, BLOCKCHAIN, COMPUTER AND INFORMATION TECHNOLOGY
(2018)
Proceedings Paper
Computer Science, Hardware & Architecture
Xiaoyuan Liang, Guiling Wang
2017 IEEE 14TH INTERNATIONAL CONFERENCE ON MOBILE AD HOC AND SENSOR SYSTEMS (MASS)
(2017)
Article
Computer Science, Information Systems
Jie Tian, Yi Wang, Xiaoyuan Liang, Guiling Wang, Yujun Zhang
Proceedings Paper
Computer Science, Hardware & Architecture
Xin Gao, Jie Tian, Xiaoyuan Liang, Guiling Wang
2014 23RD WIRELESS AND OPTICAL COMMUNICATION CONFERENCE (WOCC)
(2014)