Safe Track: AI-powered emergency braking system for train safety

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.