Rainbow Deep Reinforcement Learning Agent for Improved Solution of the Traffic Congestion
Authors:
- 1 December 2019
- conference paper
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 80-83
Abstract
While traffic congestion hits severely the world economy, adaptive traffic signal systems would efficiently provide potential solutions. In this paper, we propose a deep reinforcement learning system to control the signal lights in an isolated intersection. The proposed system uses a deep convolutional neural network to extract the crucial features from the environment state that is described by raw traffic information; i.e., vehicles positions, speeds, and waiting times. Besides, the system utilizes a multi-objective reward and the Rainbow agent which provides further space of enhancements to the conventional Deep Q-Networks agent. Extensive experiments illustrate that our proposed deep framework outperforms the baseline under a number of settings and traffic measures, including trip time, waiting time, fuel consumption, and stability.Keywords
This publication has 9 references indexed in Scilit:
- Rainbow: Combining Improvements in Deep Reinforcement LearningProceedings of the AAAI Conference on Artificial Intelligence, 2018
- Mastering the game of Go without human knowledgeNature, 2017
- Traffic light control using deep policy‐gradient and value‐function‐based reinforcement learningIET Intelligent Transport Systems, 2017
- Traffic signal timing via deep reinforcement learningIEEE/CAA Journal of Automatica Sinica, 2016
- Mastering the game of Go with deep neural networks and tree searchNature, 2016
- Human-level control through deep reinforcement learningNature, 2015
- Signal Control for Oversaturated Intersections Using Fuzzy Logic2008
- Reinforcement Learning for True Adaptive Traffic Signal ControlJournal of Transportation Engineering, 2003
- Learning to predict by the methods of temporal differencesMachine Learning, 1988