Traffic control system using Image processing and Machine learning

Project Title

Traffic control system using Image processing and Machine learning

Project Area of Specialization Artificial IntelligenceProject Summary

Our project aims to cater one of the most crucial problem faced by the modern world, Traffic congestion and traffic mismanagement which is currently the biggest problem faced in Pakistan. Traffic control systems are needed for efficient management of traffic due to rapid urbanization in developing countries(even developed countries). We will use a controller, a  camera to attain the necessary inputs(volume of traffic) for the computation of efficient signal timings. 

Project Objectives

The deliverables of our project are as the following:

1. Algorithm to predict the inflow of traffic on a certain exit. 
2. Algorithm to predict the most efficient timing for each exit
3. a machine Learning database which would be used to optimize the signal timings incase of hardware malfunction (camera not working, or image too noisy) and predict the signal flow timing using previous knowledge/data.
4. a real-time hardware system for the implementation of the above objectives.

For our additional deliverables:

1. Make the algorithm robust ie, work in harsh environmental condition (rain, fog)
2. use the algorithm to manage a 4-lane exit efficiently.

Project Implementation Method

The project is divided into two better parts. first is using image processing to identify the efficient signal timings and second is the machine-learning based signal timing calculation.

For the Image processing part, we will use basic image processing techniques and filters to mask the road and calculate the. It will incorporate image subtraction technique, blob analysis and various filtering techniques to remove the noise and make the image more readable for the computer. The input to our project will be a raw image on which we will apply these techniques to get density of cars on the traffic exit.

Then, we will construct an algorithm to estimate the efficient signal timings using previous data and the current data input (volume of cars).  This will eb done through basic amchine learning techniques to maintaina  database and to compare the input results and determine the efficient signal timing.

Thirdly, we'll develop a basic database system which would be used in case of any hardware malfunction. it will utilize previous knowledge to efficiently move the traffic without any discrepancies.

lastly, we will buy the suitable hardware and implement the code on a real time system. the hardware will consist of a controller (Odroid board) and a High-definition camera (1080p/720p) for the implementation fo the above mentioned algorithms.

Benefits of the Project

This project aims to efficiently improve the most common issue faced by the modern world that is Traffic congestion. Due to the rapid increase in urbanization in developing countries, the on-road traffic is increasing day by day which causes major traffic deadlocks.  Our goal is to make a traffic control system which uses image processing for the calculation of volume of traffic inflow and predict the most optimal time for a signal changing, which would try to efficiently improve the traffic flow and avoid unnecessary congestion on traffic signals. 

Technical Details of Final Deliverable

The final prototype will be a real-time system which would capture images in realtime from the video stream and use that as an input to the algorithm for detection of volumes. the design will be very basic (ie, a controller connected to a camera) but will also support heat-syncs to remedy the heating issue of the controller (if eneded). The camera will be used of a usb3.0 for faster transfer rate and easy accessability for the controller. 

Final Deliverable of the Project HW/SW integrated systemType of Industry Transportation Technologies Artificial Intelligence(AI), NeuroTechSustainable Development Goals Good Health and Well-Being for People, Industry, Innovation and InfrastructureRequired Resources

Item Name	Type	No. of Units	Per Unit Cost (in Rs)	Total (in Rs)
			Total in (Rs)	27000
controller	Equipment	1	14000	14000
camera	Equipment	1	12000	12000
overhead charges (not specific now)	Miscellaneous	1	1000	1000