Performance Analysis of V2X-based Real-time Vehicle Position Estimation Methods by using a Driving

초록

Researches on autonomous driving vehicles are actively conducted along with the development of various Advances Driving Assistance System (ADAS), and the development of commercially available high-level autonomous vehicles is in progress. Localization, which is to estimate the location of a vehicle, is a fundamental element of an autonomous driving where the lane-level accuracy is required for safety reason. The Global Satellite Navigation System (GNSS) is a representatively qualified method for localization of a vehicle, but there is a problem such that the code pseudorange-based single point positioning has a larger error range than a lane. Even worse, it cannot be used in the area where GNSS signals are blocked in harsh environment such as tunnels, deep urban and so on. The efficient use of Vehicle-to-Everything (V2X) is a communication between the vehicle and everything where Basic Safety Message (BSM) in the inter-vehicle communication messages of Vehicle-to-Vehicle (V2V) together with distance and angle measurements of ADAS sensors can be ideally used in solving this issue. For this, this paper presents the method to estimate a vehicle’s position in GNSS disabled environments by the efficient use of V2X and Lidar measurements, and then verifies its performance in GTA V-based driving simulator.