Intelligent Traffic Surveillance at Crossroads
Duration: January - April 2014
Status: Completed
Members: Pranav Sodhani, Akshat Bordia, Dr. Suresh Sundaram
Overview
Soaring urbanization rate and release of affordable cars in the market has undoubtedly made many a lives comfortable but has augmented traffic flow in cities by a great margin, thereby leading to frequent encounter with traffic jams. The conventional traffic light system employs fixed time operation for the traffic signals to display green/ red lights regardless of the traffic conditions. This may look reasonable at crossroads which are heavily congested but in case a lightly trafficked side road is included within the sequence it is very wasteful if in some cycles there are no vehicles waiting as the time could be better allocated to a busier approach. This project aims to develop a real-time vision system to enable smart regulation of traffic light signals by processing image frames (top view) from pole mounted video-cameras at urban crossroads. Vehicle density on each road is computed and employed as a metric to make decisions on traffic flow directions. This information is saved in a file and communicated/transmitted using cloud storage services (eg. Dropbox) to a central authority. The operator at the traffic management center owns the right to disable automated traffic signal control by editing the file as and when required. The edited file gets downloaded automatically and is processed by a microcontroller to produce the desired traffic signals. The traffic information can be published on the internet or communicated over emails/SMS to report traffic conditions throughout the city to the citizens. The system is scalable and compliant with respect to road geometry and different locations.