Abstract:
Railway transport plays a major role in daily travel and goods transportation, but accidents due
to delayed braking remain a serious issue. Traditional braking systems depend on manual operation
or fixed distance limits, which are not always reliable when speed or track conditions
change. To address this problem, this research aimed to develop an AI-powered emergency
braking system. The system enhances train safety by using low-cost electronic components
and advanced IoT technologies. The system is built using two ESP32 microcontroller boards.
The ESP32 NodeMCU is connected to an ultrasonic distance sensor to measure the real-time
distance to obstacles and a wheel speed sensor to measure the rotational speed of the train
wheels. These measurement values are sent to the Flask API hosted on Replit, where a trained
machine learning model predicts the risk of collision. If the model detects a high-risk situation
the Replit immediately sends a stop signal to the NodeMCU. NodeMCU sends that stop
signal through UART serial communication to the ESP32 DevKit V1. This board controls the
train motors through an L298N motor driver and activates an emergency stop when the stop
signal is received. In addition to the braking function, a Blynk IoT dashboard was created to
monitor the train’s real time status. The dashboard displays four main parameters: distance
from obstacles, train wheel speed in RPM, calculated risk percentage, and brake activation
status. This enables continuous monitoring of the system’s performance and safety status in
real time. Prototype tests show that the AI model provides more reliable braking decisions than
basic threshold-based approaches, as it evaluates both distance and speed before making predictions.
The project demonstrates that the use of artificial intelligence, IoT, and embedded systems
can be integrated to develop a smart, cost-effective, and scalable braking solution suitable for
real-world applications. Even though this study was conducted with prototype evaluation, the
designed mechanism will be further analyzed in real-time applications to understand potential
system mishaps before full implementation The completed system proves that applying AI in
railway safety can reduce accidents and create safer train journeys in the future.
