An IoT-based intelligent evacuation system for enhancing safety in university libraries

Abstract

The increasing demand for improved safety protocols in public spaces, such as libraries, has necessitated the development of intelligent evacuation systems that leverage the power of IoT and machine learning. This research addresses the inadequacies of traditional evacuation protocols which rely on static signage and manual alarms by proposing a dynamic system capable of real-time decision-making and adaptive guidance during emergencies. The objective of the study was to design, implement, and evaluate an IoT-based intelligent evacuation system specifically tailored for library environments. The system integrates real-time data collection through Arduino-based sensors (temperature, smoke, gas, motion, and sound), a Firebase backend for data storage and alert management, and smart cameras for environmental monitoring. A Long Short-Term Memory (LSTM) neural network was employed to predict user movement based on the "Natural Disaster Human Mobility" dataset. The system achieved a 92% accuracy rate in predicting user movement patterns and demonstrated significant improvements in evacuation efficiency during simulated emergency scenarios. These results validate the effectiveness of combining IoT technologies with machine learning for hazard detection, dynamic route optimization, and automated evacuation assistance. This research concludes that the proposed intelligent system enhances public safety by minimizing response time and providing data-driven evacuation instructions, offering a scalable model for adoption in similar public environments. Keywords: Intelligent evacuation systems, IoT, libraries, machine learning, real-time data, public safety, emergency management, dynamic evacuation routes.