MAP Aided Self-positioning Based on LIDAR Perception for Autonomous Vehicles

Abstract

Precise localization is a necessity for complex driving maneuvers of autonomous vehicles. This paper presents a high-definition-map aided self-positioning for autonomous driving. Sparse landmarks, including shuttle stops, speed limit signs, and guide signs, are recorded in the map. A 16-layer light detection and ranging device is used to detect the selected landmarks. Based on the distance and orientation information of the landmarks, the proposed method can be applied to estimate the vehicle’s heading and location. Simulation results show that the 67% of heading and location errors are below 2.2 degrees and 30 cm, respectively.