Rail Tracking and Detection with Drone in Gazebo Environment

Abstract

Rail transport is considered one of the safest modes of transport. Along with the development of technology, significant developments have been observed in railway transportation over the years. With the increasing railway line, the demand for railway transportation is increasing day by day. The number of passengers using railway transportation has also increased in this context. With this intensity, the damage to the railway line increases. In ensuring the safety of railway transport, methods based on deep learning have become important to ensure railway safety. In order for the railway line to provide healthy service, the monitoring of the railway line should be done regularly. Traditional rail monitoring services require different vehicles. Today, besides conventional vehicles, drones are used for the monitoring of the railway line. Experimenting with drones in the real environment can be difficult and costly. It is always more advantageous to run the codes that will run on the drone first in a simulation environment to save time. In this study, a drone-based system that autonomously tracks and detects railway tracks is proposed as an alternative to traditional methods. The proposed method detects the rails with the semantic segmentation method and follows the rails with its front camera. The proposed method was developed in the Gazebo environment. The general purpose of the study is to record the rail images with the drone camera that follows the railway autonomously. In this way, drone experiments to be carried out in the real railway environment will be completed in a shorter time.