Background
Urban traffic became more and more crowded in the past years due to the increasing number of cars and pedestrians that traffic congestion is non-linear to the rapid development. Traditional traffic lights blink the light signals after a certain time period under the premier investigation of traffic flow in which way cannot solve the traffic problems fully (Maheswari, et.al., 2018 ). In fact, the change of vehicle flow is usually uncertain and it may differ a lot among different intersections and different times. This phenomenon would occur: There are few vehicles in the green direction while there are long queues in the red direction waiting to pass. In order to solve the traffic flow, option available is to detect the pedestrians and vehicles based on smart smart traffic control. Smart traffic lights are a vehicle traffic control system under artificial intelligence to monitor and manage vehicle and pedestrian traffic in real time (wikipedia). From technical perspective, the system is based on computers and sensor networks.
Problem
The main stumbling difficulties to the application of smart traffic control system is that the pedestrian and vehicle flow could not communicate with the computer systems well that the smart system could not work well to control traffic.
Aims
It will propose the machine learning algorithms which can be used by the systems for detection, analysis and congestion prediction. The goal is that when the sensor network goes on real-time monitoring of different pedestrians and variety kinds of vehicles, e.g. bicycles, cars, trucks, the algorithms could help system to learn how to make a judgement to make appropriate adjustment and learn to predict the congestion which is about to occur. Thus, the system can control the traffic automatically, release the traffic congestion and reduce the congestion time, especially during the peak period. It would make use of Congestion Algorithm proposed by Suguna Devi et.al as figure shows (Devi and Neetha, 2017).
It would also carry out experimentation using different configuration of the test data and analyse the data of performed experimentation to achieve a better algorithm of the traffic control system.
Hardware and Software Requirements
As to the hardware perspective, it needs a normal computer that provides enough programming environment.
The software of whole system would adopt Python (the main programming tool), PYCharm (the tool to help increase the efficiency) and Tensorflow (mainly used for realisation of machine learning algorithm) to programme.
Plan
•Initial Literature Review (4 weeks)
◊Broad reading about the existed research and design related to the smart traffic control system
◊Identify the proposal and issues
•Analysis (5 weeks)
◊Analysis of existed research, design and applications
◊Identify the problems, aims and methodology
◊Define the hardware and software requirements
•Design Process (13 weeks)
◊Explore and produce
◊Design for proposed smart traffic system
◊Testing of the system (Experimentation)
•Implementation and Evaluation (6 weeks)
◊Implement the framework
◊Evaluate the designed system
◊Complete the report
Gantt Chart
The following Gantt chart has given the main stage of the proposal plan and the start and end time. The actual time could be edited after the project is completed and it can be compared with the schedule. It can also make out the percentage of the work that has been carried out and find out the quality of the work.
References
• Maheswari, RA. Preethi, R. Poorvaja, A. Mahizhini, P.R. Preethi. (2018). DETECTION OF TRAFFIC CONGESTION USING SMART TRAFFIC CONTROL SYSTEM. IJARIIE-ISSN(O)-2395-4396.
• Suguna Devi and T. Neetha. (2017). A Novel Algorithm for predicting road traffic congestion in a IOT based smart city. IIERD. Special Issue SIEICON-April, p-ISSN: 2348-6406.