Vehicle Detection for Collision Avoidance Using Vision based Approach: A Constructive Review

Abstract

Computer vision in the context of automation has been commonplace in industrial manufacturing for decades. A new wave of opportunity driven to achieve autonomous capability for vehicles to avoid collision is a potential demand for the Fourth Industrial Revolution (IR 4.0). In this context, vehicle detection for collision avoidance has started to get attention in recent years to reduce vehicle collision during vehicle overtaking. In collision avoidance, main aims of vehicle detection are to alert the driver about driving environment and avoid collision with other vehicles which demands for efficient vehicle detection methodology. This research demonstrates the analysis about detail difficulties for vehicle detection in terms with overall procedures. To accomplish efficient vehicle detection methodology, vehicle detection for collision avoidance is illustrated with three steps which involves hypothesis generation, hypothesis verification and finally, vehicle tracking. Each of the steps is demonstrated with adequate methods with advantages and disadvantages which make it easier to choose best method in each step. The overall reviews proposed in this research have been extensively studied in previous research is expected to fulfill the future demand of autonomous vehicles for Fourth Industrial Revolution (IR 4.0).