Description:
Field of the Invention:

The present invention relates to railway safety systems, and provides real-time alerts of approaching trains using AI-based computer vision, radar sensing, thermal imaging and M2M SIM technology. The system is designed to enhance safety at unmanned railway crossings, blind curves, deep forests and other accident-prone zones.

Background of the Invention

At the root, the invention stemmed from the need to ensure safety of factory workers loading and unloading urea bags onto the goods carriage. There are 5479 virtual junctions in India today, where the solution is very much required. Virtual junctions are industry offtakes on the rail where freight carrying trains reach the industry for loading and unloading. This particular invention was done at the Matix offtake at Panagarh Industrial Estate, where goods trains do not have a scheduled arrival / departure and the life of employees crossing the track or working on the platform were at risk.

Looking beyond this specific case, railway crossings, especially unmanned and remote ones, are often the site of serious accidents involving trains and road vehicles or pedestrians. Traditional safety mechanisms such as flagmen, static hooters, and track-based sensors may not be deployed always or may not be reliable due to maintenance challenges, environmental conditions, and human errors.

Furthermore, most existing systems cannot detect in heavy rain or smog/fog due to poor visibility. Overall, there is a lack of cost-effective, scalable, and autonomous systems that can issue timely alerts in real-world, uncontrolled environments.
There is, therefore, a need for a robust, intelligent, and automated alert system capable of operating autonomously and delivering timely warnings at critical railway zones to prevent accidents and enhance public and operational safety.

Such vehicle detection cum alert systems are equally applicable to any form of transport be it by rail or by road or by waterway or in the air.

Summary
The present invention uses ubiquitous sensor technologies like the camera and the radar to detect the train even in harsh weather with heavy fog and rain, or at night and send the data in remote areas using M2M SIM to a MQTT broker topic. The same topic is subscribed to at the alert end where a set of sirens and flashers become on to announce the approach of a train and alert the workers on the platform. It does a similar thing as the train moves in the opposite direction alerting the workers at the level crossing or on the platform.
The invention comprises:
1. A Raspberry Pi 5 edge computing unit capable of processing data in real-time.
2. A camera module configured to capture video feed of the railway track
3. An AI-based object detection algorithm such as YOLO (You Only Look Once) v8 to identify approaching trains and disregard JCBs, bikes, trucks and
4. A radar sensor positioned alongside the track to detect moving objects, providing redundancy and ensuring detection during adverse conditions such as fog, dust, rain, or poor lighting.
5. A communication module utilizing MQTT, and M2M SIM to publish messages if a train is detected in real time to the MQTT broker on Plumule cloud
6. A hooter or siren system triggered by the detection of a train to warn nearby pedestrians or vehicles.
7. Flashing lights or visual indicators activated along with the hooter to increase visibility of the alert.

8. The system is configured to operate autonomously, without requiring human intervention, and is suitable for deployment in remote and unmanned locations. The edge processing ensures low-latency response, while the dual-mode sensing (vision + radar) improves accuracy and reliability.
, C , C , Claims:An unmanned train detection and alert system, comprising:
1. A train detection module, a processing unit, and an alert module, wherein the system autonomously detects the presence of an approaching train and triggers a corresponding alert without requiring manual intervention.

2. The system as claimed in claim 1, wherein the train detection module comprises one or more of the following technologies: radar, image processing via camera module.
3. The system as claimed in claim 1 or 2, wherein the processing unit includes embedded computing board configured to receive signals from the detection module and determine train proximity or speed.
4. The system as claimed in any preceding claim, wherein the alert module comprises at least one of the following: a high-decibel hooter, buzzer and wireless alert system for remote monitoring using relay .
5. The system as claimed in any preceding claim, wherein the alert module is activated when the train is detected within a predefined detection zone or based on the train’s approach velocity.
6. The system as claimed in any preceding claim, further comprising a communication interface configured to transmit real-time train detection data to a central monitoring station or mobile application via GSM Wi-Fi.
7. The system as claimed in any preceding claim, wherein the system operates continuously in all environmental conditions including rain, fog, dust, and night-time using robust sensor calibration and weatherproof enclosures.
8. The system as claimed in any preceding claim, wherein the processing unit is further configured to log train passage events with timestamp and store or upload the data for audit and analytics purposes.
9. The system as claimed in any preceding claim, wherein multiple detection modules can be deployed across various distances to enable redundancy, increased accuracy, and longer detection range.