An Integrated Eco-Aware Rail Management System

This paper focuses on the development of a novel and advanced Eco-Aware Rail Management System that would improve railway safety using multi-sensor fusion. This system detects and tracks potential obstacles, along the rail tracks and guides the control cabin. The method, intelligent obstacle detection system based on deep learning employs sophisticated deep learning techniques, specifically leveraging Convolutional Neural Networks (CNNs), to process data from cameras and LiDAR sensors. By employing semantic segmentation and object detection algorithms, this method achieves pixel-wise identification of railway tracks and accurate detection of potential obstacles. The fusion of camera and LiDAR data enhances environmental perception across varied weather conditions, enabling precise determination of geometric relationships between tracks and obstacles. With GPS and INS data, braking/deceleration response is generated based on the state of the control cabin.