Train Frontal Obstacle Detection Method with Camera-LiDAR Fusion

Abstract

Recently, the importance of obstacle detection methods for railway has been increasing. In the field of automobiles, obstacle detection systems with sensors have been introduced on mass-produced vehicles. However, in railway, a practical detection system does not exist because railways require longer detection distances than do automobiles. Therefore, we have developed a train frontal obstacle detection method using a camera and LiDAR. We confirmed that our method detects a person 200 m away, which a camera alone cannot detect, with 45% accuracy at night.