Cooperative LiDAR-Aided Self-localization of CAVs in Real Urban Scenario

In the rapidly advancing realm of Connected Autonomous Vehicles (CAVs), achieving reliable and precise positioning is of paramount importance. This paper presents a comprehensive approach integrating vehicular sensing, communication, and advanced filtering techniques to enhance vehicle positioning in urban areas. By leveraging LiDAR point clouds along with a light and accurate object detector, we create cohesive environmental sensing that improves situational awareness in autonomous systems. Central to our methodology is the integration of the Labeled Multi-Bernoulli Mixture (LMBM) filter, which offers a probabilistic framework for dynamic state estimation in environments characterized by high uncertainty and clutter. In turn, enhanced object locations are exploited as anchors for vehicular self-localization via an Extended Kalman filter (EKF). Our experimental results show that the proposed solution significantly enhances vehicular positioning accuracy.