Urban Rail Transit Track Area Extraction Using LiDAR

The intrusion of obstacles into the track area is a significant issue that impacts the safety of urban rail transit. The obstacle detection technology based on LiDAR (LIght Detection and Ranging) meets the detection requirements of urban rail transit. However, under the limitations of current LiDAR performance, existing methods can only extract the track area up to 100 m, which is insufficient for the speed and braking distance of urban rail transit trains. In light of the characteristics of urban rail transit lines, a new method is proposed to indirectly extract the track area by detecting references parallel to the rail, i.e., tunnel walls, protective walls, and sound barriers. By means of border points selection, clustering, and curve fitting, two curves of the references on both sides are obtained. A track centerline is obtained from the two curves and is expanded to extract the track area according to size specifications. The proposed method in this paper is evaluated in a variety of scenes, including tunnel, elevated, and ground sections of urban rail transit lines. The result shows that the proposed method successfully extracts the track area within 150 m.