Coordinated Deep Reinforcement Learners for Traffic Light Control

By van der D David Pol. 2016

How researchers see this paper

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

In Reinforcement Learning

Citations in the corpus

Value-Decomposition Networks For Cooperative Multi-Agent Learning (2017, methodology): “Our approach works with only a team reward, and learns the value-decomposition autonomously from experience, and it similarly differs from the approach with coordination graphs (Guestrin et al., 2002) and the max-plus algorithm (Kuyer et al., 2008; van der Pol and Oliehoek, 2016).”
Breaking the Curse of Horizon: Infinite-Horizon Off-Policy Estimation (2018, methodology): “We consider the task of reducing traffic congestion by modelling traffic light control as a reinforcement learning problem [41].”
Stabilising Experience Replay for Deep Multi-Agent Reinforcement Learning (2017, background): “However, many real-world problems, such as network packet delivery (Ye et al., 2015), rubbish removal (Makar et al., 2001), and urban trafﬁc control (Kuyer et al., 2008; Van der Pol & Oliehoek, 2016), are naturally modeled as cooperative multi-agent systems.”
The Representational Capacity of Action-Value Networks for Multi-Agent Reinforcement Learning (2019, background): “In this paper, we focus on value-based multi-agent reinforcement learning (MARL) [2, 5, 6, 14, 20, 24] approaches for cooperative multi-agent systems (MASs) [10, 12, 22, 26, 29].”
A Study on Overfitting in Deep Reinforcement Learning (2018, background): “…and fundamental infrastructures such as power grids (Wen et al., 2015; Glavic et al., 2017) and traffic control (Mannion et al., 2016; van der Pol & Oliehoek, 2016), it is becoming crucial to establish good understandings of the representation learning, long term planning, exploration and…”
A survey and critique of multiagent deep reinforcement learning (2018, background): “However, note that other works have shown batch normalization can cause divergence in DRL [274,335].”
Value-Decomposition Networks For Cooperative Multi-Agent Learning Based On Team Reward (2018, background): “[10] and the max-plus algorithm [13, 26] also rely on specified individual rewards.”
Multi-Agent Deep Reinforcement Learning for Large-Scale Traffic Signal Control (2019, methodology): “Reference [37] applied deep IQL, with the partial observability addressed with transfer planning, but the states were infeasible and the simulated traffic environments were oversimplified in that study.”
IntelliLight: A Reinforcement Learning Approach for Intelligent Traffic Light Control (2018, methodology): “To solve the in-managablely large state space of previous methods, recent studies [15, 22] propose to apply Deep Q-learning methods using continuous state representations.”
CoLight: Learning Network-level Cooperation for Traffic Signal Control (2019, background): “For example, studies [16, 24] have proposed to jointly model two adjacent intersections.”
CityFlow: A Multi-Agent Reinforcement Learning Environment for Large Scale City Traffic Scenario (2019, background): “Traffic signal control problem, one of the biggest urban problems, is drawing increasing attention in recent years [5, 9, 10].”
Multi-objective multi-agent decision making: a utility-based analysis and survey (2019, background): “Finally, one can identify smaller components inside a complex setting, solve each task individually, and then use transfer planning [113] to go back to the original problem, mitigating issues such as non-stationarity [121].”
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.”
Toward A Thousand Lights: Decentralized Deep Reinforcement Learning for Large-Scale Traffic Signal Control (2020, methodology): “To achieve coordination, an alternative way is through centralized optimization over multiple coordinated agents in an area to ensure the optimality (Kuyer et al. 2008; Van der Pol and Oliehoek 2016; El-Tantawy, Abdulhai, and Abdel-gawad 2013).”