Real-Time Vision-Based & Optimized Pedestrian Crossing Control System

Abstract

As per the National Council for Road Safety, more than 900 fatalities have been reported due to vehicle collisions on or near pedestrian crossings during the past 10 years within Sri Lanka. However, only minimal research has been carried out so far to improve the safety conditions of pedestrians. This research focuses on developing a system utilizing image processing to obtain the real-time data of a group of pedestrians at crosswalks and determining of an optimal time frame for a certain group of pedestrians to cross the road at a given time period. This study concludes on the necessary requirements to optimize the timing of the existing pushbutton based signaling system. A mobile based app is also developed to assist drivers to be incorporated with the proposed signaling system to enhance the end-effect. Extensive trials prove that the proposed system would ensure pedestrian safety and minimize or even prevent reckless road collisions within crosswalks.