Automated Railway Crossing and Obstacle Detection
7 Pages Posted: 21 Apr 2019
Date Written: April 8, 2019
The aim of this project is to automate the level crossing using a micro-controller and object detection system to detect objects on the track at the crossing. One of the main problems in railway transportation is the level crossings gate control system. Currently in Indian railways the gate is controlled by a gatekeeper. Whenever the train arrives, that information will be sent to the gatekeeper. Based on that information the gate will be closed and opened by him. If there is any irresponsibility by the gatekeeper, then there will be a big problem. This can be avoided by using an automated railway gate control system which uses sensors that detects the arrival and departure of the train and closes and opens the gates accordingly. Under normal operating condition, when the first sensor is triggered by the train which is approaching the gates, the gates are closed, and when the second sensor is triggered by the train the gates are opened. A camera is installed near the level crossing, which acts as the input video stream, for an object detection algorithm, the object detection is performed in real-time. Thus, if after the first sensor is triggered and then if a pedestrian or a vehicle is detected between the tracks, by this object detection algorithm, then the gates are automatically reopened, to let the stuck vehicle or pedestrian cross. At the same time a signal is sent to the train, to indicate that a vehicle or pedestrian is on the tracks. The gates remain open until the object is detected, as soon as the stuck person or vehicle crosses the tracks, the gates are closed again and a signal is sent to the train, indicating that the tracks are clear. This is done to prevent the subsequent pedestrian or driver from thinking that the train is not in the vicinity. The object detection data is sent in real-time to firebase database, over the internet, and it can be viewed from anywhere using a mobile phone or a computer.
Keywords: Arduino, Nodemcu, TensorFlow, IOT, Machine Learning, Real-time Object Detection, Firebase Database, Python
Suggested Citation: Suggested Citation