Regional risk estimation for drivers cutting intelligent graph with intra cells enabling risk transfer for street players

Abstract

As a vehicle proceeds on urban road that includes possible street players such as pedestrian, vehicles, traffic signs, traffic lights, or crosswalks, we can mention about holistic risk caused by dynamic behavior of each items moving along stochastic directions. In the numerous studies, traditional risk estimation methods have been devoted to utilize traffic agents, which have not resulted in effective outcome for a vehicle resided at somewhere among these agents and sometime at its own place. In our scenario, we consider total emerging risks for a driver owing to continuous movements of some of agents on the scene. Specified partitions are envisaged for each agents referring to players at the sight of driver, each of which might have risks as the time goes on. One or multiple agents may occupy in same partition. Therefore, we can mention about potential risks for the subject vehicle. This study has inspired from conventional Graph Cut theory. Graph vertex representing partitions may have transferable risk capacity for the agents in the scene. Vertexes can be imitated to memory cell array that have transferable risk capacities enabling the system to estimate instant risks as traveling in city traffic. As an outcome of this study, the proposed system suggests a driver to take a caution while moving along the street at clutter environment.