Enhanced Vehicle Positioning in Cooperative ITS by Joint Sensing of Passive Features

Abstract

Satellite-based navigation systems, such as Global Positioning System (GPS) or Galileo, are the most common and accessible techniques for vehicle positioning. However, in dense urban areas, even if combined with vehicle on-board sensors, they lead to large localization errors due to multipath and signal blockage. In recent years, cooperative intelligent transportation systems (C-ITS) have gained increasing attention as they allow vehicles to cooperate and broadcast safety-related information to the neighbors through vehicle-to-vehicle (V2V) communications. In this paper, a novel cooperative positioning method is developed by exploiting V2V communications without using explicit V2V ranging. Vehicles localize, in a distributed way, a set of jointly sensed non-cooperative features (e.g., people, traffic lights) and use them as common noisy reference points to implicitly enhance their own position accuracy. Distributed belief propagation is combined with consensus-based estimation of the features' positions to enable cooperative localization of vehicles. Numerical results show that the proposed method is able to significantly enhance the GPS-based vehicle location accuracy, reducing the error by approximately a factor 2 in the considered scenario, or even more in setups with denser feature deployments.