Smart Traffic Management System for Smart Cities using Reinforcement Learning Algorithm
D. Venkata Siva Reddy1, R.Vasanth Kumar Mehta2
1D. Venkata Siva Reddy, Research Scholar, Department of Computer Science & Engineering, SCSVMV University, Kanchipuram (Tamil Nadu), India.
2Dr. R.Vasanth Kumar Mehta, Associate Professor, Department of Computer Science & Engineering, SCSVMV University, Kanchipuram (Tamil Nadu), India.
Manuscript received on 21 March 2019 | Revised Manuscript received on 02 April 2019 | Manuscript Published on 18 April 2019 | PP: 12-15 | Volume-7 Issue-6S March 2019 | Retrieval Number: F02060376S19/2019©BEIESP
Open Access | Editorial and Publishing Policies | Cite | Mendeley | Indexing and Abstracting
© The Authors. Blue Eyes Intelligence Engineering and Sciences Publication (BEIESP). This is an open access article under the CC-BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/)
Abstract: Traffic congestion at junctions or on roads may be seen due to many reasons like slow driving, increased vehicle queue etc. In some emergency cases when vehicles are stopped for longer period traffic jam may occur. Adaptive traffic control system is a traffic management strategy used to control the traffic by facilitating the signals to instantaneously adjust to the present traffic demand. Adaptive traffic signal functions by utilizing both hardware and software coordination. Q-learning needs already designed precise form of the environment for selecting action. As a substitute, we can adopt dynamic communication system to find the interaction between state, action and rewards of that particular environment. The present traffic signal works based on the pre specified traffic flow data to extract short time anticipation which helps to evaluate the consequence on the signal controlling system. If single model is used then at each and every time adaptive traffic light control agents need to collect photographs of the existing condition of the traffic and generate control signals. We have implemented occurrence, replay and ideal mechanisms to improve the consistency of the algorithm. The main aim in designing the algorithm is to control overcrowded traffic and for this we have incorporated the dynamic network with the linear signal arrangement.
Keywords: Traffic Control, Reinforcement Learning, Deep Learning, Value-function Method and Artificial Neural Networks.
Scope of the Article: Algorithm Engineering