ISSN 2394-5125
 

Research Article 


DEVELOPMENT OF AN OPTIMAL ROAD-TRAFFIC CONTROLLER THROUGH REINFORCEMENT LEARNING

RUDOLPH JOSHUA CANDARE, ADRIAN PETE LAGARE, ELLY JOHN MIRASOL, JAYMER JAYOMA, ROLYN DAGUIL.

Abstract
The common way to manage road intersection traffic is done through traffic lights. Existing traffic light intersection controllers have a fixed timing
in their outputs without considering the real-time traffic situation. This study aims to develop a smart traffic light controller which is adaptive
depending on the traffic situation in order to lessen the traffic in intersections. In order to achieve the smart traffic light controller, reinforcement
learning is the technique in which the researchers implement for its development. Reinforcement Learning (RL) is a type of machine learning
technique that best fits this problem. It learns by interacting through its environment and given rewards and punishment based on its actions. In
this paper, the researchers used two RL methods, namely Deep Q-Learning/Network (DQN) and Covariance Matrix Adaptation EvolutionaryStrategies (CMA-ES) with the same deep neural network architecture to implement the smart traffic light controller and will be simulated via Unity
3D. The smart traffic light agent is tested in two test areas with easy and hard levels of traffic with DQN, CMA-ES together with the Fixed-Time for its
basis. The results of this study demonstrate the advantage of using two different RL methods over a fixed time traffic controller. The researchers
found out that CMA-ES had better performance for a smart traffic light controller.

Key words: Reinforcement Learning, Traffic Control, Deep Q-Learning, Covariance Matrix Adaptation Evolutionary-Strategies, Deep Neural Network


 
ARTICLE TOOLS
Abstract
PDF Fulltext
How to cite this articleHow to cite this article
Citation Tools
Related Records
 Articles by RUDOLPH JOSHUA CANDARE
Articles by ADRIAN PETE LAGARE
Articles by ELLY JOHN MIRASOL
Articles by JAYMER JAYOMA
Articles by ROLYN DAGUIL
on Google
on Google Scholar


How to Cite this Article
Pubmed Style

RUDOLPH JOSHUA CANDARE, ADRIAN PETE LAGARE, ELLY JOHN MIRASOL, JAYMER JAYOMA, ROLYN DAGUIL. DEVELOPMENT OF AN OPTIMAL ROAD-TRAFFIC CONTROLLER THROUGH REINFORCEMENT LEARNING. JCR. 2020; 7(15): 612-615. doi:10.31838/jcr.07.15.92


Web Style

RUDOLPH JOSHUA CANDARE, ADRIAN PETE LAGARE, ELLY JOHN MIRASOL, JAYMER JAYOMA, ROLYN DAGUIL. DEVELOPMENT OF AN OPTIMAL ROAD-TRAFFIC CONTROLLER THROUGH REINFORCEMENT LEARNING. http://www.jcreview.com/?mno=119826 [Access: September 14, 2020]. doi:10.31838/jcr.07.15.92


AMA (American Medical Association) Style

RUDOLPH JOSHUA CANDARE, ADRIAN PETE LAGARE, ELLY JOHN MIRASOL, JAYMER JAYOMA, ROLYN DAGUIL. DEVELOPMENT OF AN OPTIMAL ROAD-TRAFFIC CONTROLLER THROUGH REINFORCEMENT LEARNING. JCR. 2020; 7(15): 612-615. doi:10.31838/jcr.07.15.92



Vancouver/ICMJE Style

RUDOLPH JOSHUA CANDARE, ADRIAN PETE LAGARE, ELLY JOHN MIRASOL, JAYMER JAYOMA, ROLYN DAGUIL. DEVELOPMENT OF AN OPTIMAL ROAD-TRAFFIC CONTROLLER THROUGH REINFORCEMENT LEARNING. JCR. (2020), [cited September 14, 2020]; 7(15): 612-615. doi:10.31838/jcr.07.15.92



Harvard Style

RUDOLPH JOSHUA CANDARE, ADRIAN PETE LAGARE, ELLY JOHN MIRASOL, JAYMER JAYOMA, ROLYN DAGUIL (2020) DEVELOPMENT OF AN OPTIMAL ROAD-TRAFFIC CONTROLLER THROUGH REINFORCEMENT LEARNING. JCR, 7 (15), 612-615. doi:10.31838/jcr.07.15.92



Turabian Style

RUDOLPH JOSHUA CANDARE, ADRIAN PETE LAGARE, ELLY JOHN MIRASOL, JAYMER JAYOMA, ROLYN DAGUIL. 2020. DEVELOPMENT OF AN OPTIMAL ROAD-TRAFFIC CONTROLLER THROUGH REINFORCEMENT LEARNING. Journal of Critical Reviews, 7 (15), 612-615. doi:10.31838/jcr.07.15.92



Chicago Style

RUDOLPH JOSHUA CANDARE, ADRIAN PETE LAGARE, ELLY JOHN MIRASOL, JAYMER JAYOMA, ROLYN DAGUIL. "DEVELOPMENT OF AN OPTIMAL ROAD-TRAFFIC CONTROLLER THROUGH REINFORCEMENT LEARNING." Journal of Critical Reviews 7 (2020), 612-615. doi:10.31838/jcr.07.15.92



MLA (The Modern Language Association) Style

RUDOLPH JOSHUA CANDARE, ADRIAN PETE LAGARE, ELLY JOHN MIRASOL, JAYMER JAYOMA, ROLYN DAGUIL. "DEVELOPMENT OF AN OPTIMAL ROAD-TRAFFIC CONTROLLER THROUGH REINFORCEMENT LEARNING." Journal of Critical Reviews 7.15 (2020), 612-615. Print. doi:10.31838/jcr.07.15.92



APA (American Psychological Association) Style

RUDOLPH JOSHUA CANDARE, ADRIAN PETE LAGARE, ELLY JOHN MIRASOL, JAYMER JAYOMA, ROLYN DAGUIL (2020) DEVELOPMENT OF AN OPTIMAL ROAD-TRAFFIC CONTROLLER THROUGH REINFORCEMENT LEARNING. Journal of Critical Reviews, 7 (15), 612-615. doi:10.31838/jcr.07.15.92