Estimating time to interaction for vehicles in ITS applications

Abstract

The detection and avoidance of vehicle collisions is an important area of research in the field of Intelligent Transportation Systems (ITS). Time To Collision (TTC) is widely used as an indicator of impending collisions between two vehicles. This is generally calculated using sensors and/or V2V communication to determine the state of nearby vehicles, and by performing a future projection of the state to estimate a collision. This paper introduces a technique using a digital map to incorporate features of the road environment into the vehicle model, providing constraints on the potential vehicle motion. This model provides an estimate of the 'real' minimum distance between vehicles given the road environment, the likely mode of interaction, and an estimate of the Time To Interaction (TTI) between a pair of vehicles. The metrics provide a high level estimate regarding the potential for future vehicle interactions, which can be augmented by incorporating additional measures of estimated driver intention. The metrics can also be incorporated into low level collision avoidance systems as a filter for vehicles that should not be considered a threat due to map constraints, which reduces the likelihood of false positives.