A New Approach to Positioning Error Mitigation: How to Learn Via Safety Messages

Accurate positioning of vehicles and vulnerable road users is critical for ensuring road safety. Today's localization relies on the Global Navigation Satellite System (GNSS), but it is often compromised by signal obstructions due to urban infrastructures, buildings, and other vehicles. To overcome these limitations and enhance GNSS localization accuracy, a cooperative positioning method is put forth that leverages direct vehicular communications and the onboard sensors of the vehicle. In the proposed approach, a real-time algorithm selects the vehicle with the lowest positioning error as the anchor node and mandates the anchor to broadcast its coordinates via a suitable message. The receiving vehicles whose sensing region includes the anchor can determine the distance from it and update their position estimate accordingly. In turn, these vehicles may serve as secondary anchors, extending the correction process to vehicles beyond the first tier. By jointly exploiting direct vehicular communications and the local estimate of the distance from the anchor, the proposed approach achieves accurate vehicle localization. In a reference urban intersection, taking into account imperfect direct vehicle-to-vehicle communications and the first tier only, the percentage of vehicles improving the estimate of their position lies between 36% and 54%, depending on the penetration rate of the connected vehicles; these values raise to 62% and 88% when the secondary anchors contribute to propagate the error correction. Furthermore, the statistical distribution of the positioning error exhibits a significant shift toward lower error values, raising the probability that the error is lower than 6 m from 0.05 when no correction is introduced, to 0.87 when all the vehicles are connected and imperfect communications are considered.