Reinforcement learning-based multi-agent system for network traffic signal control

By Itamar Arel, C. Liu, T. Urbanik, A. Kohls. 2010

520 citations recorded by the source corpus.

Subfields: Building and Construction, Control and Systems Engineering, Transportation

View Paper: https://www.semanticscholar.org/paper/8bc80ccc97902264e9d98c48f7ad36f7dcd61b3c

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Building and Construction

In Control and Systems Engineering

In Transportation

Citations in the corpus

Explainability in Deep Reinforcement Learning (2020, background): “Notable examples include training agents to play Atari games based on raw pixels [3, 4], board games [5, 6], complex real-world robotics problems such as manipulation [7] or grasping [8] and other real-world applications such as resource management in computer clusters [9], network traffic signal co”
A review of cooperative multi-agent deep reinforcement learning (2019, background): “Another approach is to assume that agents could share their states among the neighbors [8].”
Integrating Independent and Centralized Multi-agent Reinforcement Learning for Traffic Signal Network Optimization (2019, background): “On the other hand, existing work that adopt the multi-agent perspective on traffic signal optimization either fall back to independent learning [16] or resort to centralized optimization of coordinated agents [2, 30].”
Toward A Thousand Lights: Decentralized Deep Reinforcement Learning for Large-Scale Traffic Signal Control (2020, background): “• NeighborRL (Arel et al. 2010): a multi-agent deep Qlearning algorithm that feeds the model with both its own and neighbors’ observations for network-level cooperation.”
IntelliLight: A Reinforcement Learning Approach for Intelligent Traffic Light Control (2018, background): “These works vary in the state representation including hand craft features (e.g., queue length [15, 17], average delay [10, 22]) and image features[9, 16, 22]) They are also different in reward design, including average delay [3, 22],the average travel time [16, 22], and queue length[15].”
A Deep Reinforcement Learning Network for Traffic Light Cycle Control (2018, background): “Due to the technique limitation at the time in reinforcement learning, they usually make a small-size state space, such as the number of waiting vehicles [8], [22] and the statistics of traffic flow [23], [24].”
A Survey on Traffic Signal Control Methods (2019, methodology): “,b] Waiting time [Chu et al. 2019; Wei et al. 2018] Volume [Aslani et al. 2017, 2018b; Balaji et al. 2010; Cahill et al. 2010; Casas 2017; El-Tantawy and Abdulhai 2010; Wei et al. 2019a, 2018] Delay [Arel et al. 2010] Speed [Casas 2017; El-Tantawy and Abdulhai 2010; Nishi et al. 2018] Phase duration”
Explainable Reinforcement Learning: A Survey (2020, unknown): “Possible applications for the use of RL are teaching neural networks to play games like Go [55], teaching robots to perform certain tasks [32], or intelligent transport systems [4].”
Using a Deep Reinforcement Learning Agent for Traffic Signal Control (2016, background): “Previous research efforts have defined the state space as some attribute of traffic, the number of queued vehicles [10][12][13][14] and traffic flow [15][6] the”
Big Data Analytics in Intelligent Transportation Systems: A Survey (2019, background): “One classic example of using reinforcement learning in ITS traffic signal control is formulated in [76].”
Traffic Light Control Using Deep Policy-Gradient and Value-Function Based Reinforcement Learning (2017, background): “Developing advance traffic simulation tools have made researchers develop novel state representation and reward functions for reinforcement learning algorithms, which could consider more aspects of complexity and reality of real-world traffic problems [13, 1, 8, 3].”
Graph Neural Networks for Intelligent Transportation Systems: A Survey (2023, methodology): “Finally, they compared the framework with baselines (fixed time [158], Max Pressure [159], CGRL [160], Individual RL [161], OneModel [162], Neighbor RL [163], GCN [155], and Co-Light [157]) on two real datasets, and concluded that their model outperforms the state-of-the-art.”