Sensor Fusion Method for Train Forward Surveillance

We present the development of a train forward surveillance method using a camera and sensors. In train forward surveillance methods for railroads, it is crucial to establish a sensor technology capable of reliably detecting distant obstacles hundreds of meters away. Therefore, we developed a detection method by sensor fusion between 4K camera and 3D LiDAR. Using a combination of a camera and nine LiDARs, we confirm that the proposed method can detect a person located 400m away with a detection rate of 94% and a car located 600m away with a detection rate of 100%. Moreover, based on the trend that detection performance improves as the number of LiDARs increases, we estimated the conditions required to reliably detect a person 500m away.