A Probabilistic Model for the Estimation of Pedestrian Crossing Behavior at Signalized Intersections

Abstract

Active safety systems which assess highly dynamic traffic situations including pedestrians are required with growing demands in autonomous driving and ADAS. In this paper, we focus on one of the most hazardous traffic situations: the possible collision between a pedestrian and a turning vehicle at intersections. This paper presents a probabilistic model of pedestrian behavior to signalized crosswalks. For this purpose, we take not only pedestrian physical states but also contextual information into account. We propose a model based on the Dynamic Bayesian Network (DBN) which integrates relations among the intersection context information and the pedestrian behavior in the same way as human. Afterwards, the model jointly estimates their states by the particle filter. Experimental evaluation using real traffic data shows that this model is able to recognize the pedestrian crossing decision in advance from the traffic signal and pedestrian position information.