RMF based target position estimation and collision probability forecasting technique in freeway junctions

Abstract

Collision between vehicles and pedestrians leads to brutal loss of life and assets on Indian roads. Accidents happen due to individual's negligence and misjudgement of the speed of vehicles at freeway junctions. In this paper, a novel feature extraction technique is used for estimating the target position and to update the trajectory information. A vision-based technique is incorporated to acquire the target information that is a simple and cost-effective method to examine the target's current position. Moreover, a distribution-based evaluation method is introduced to calculate the degree of conflict and avoid crashes by alerting the target. The experimental results of the proposed technique reveal an improved performance of 9% in detection rate for public datasets over the existing Gaussian mixture model (GMM) method. The proposed probabilistic collision avoidance system could be implemented on highways to reduce the accidents to a greater extent.