Crowd movement monitoring in academic buildings: a reinforcement learning approach

作     者：Geetha, T.S. Rao, C. Subba Chellaswamy, C. 

作者机构：Department of Electronics and Communication Engineering SA Engineering College Chennai600077 India Department of Electronics and Communication Engineering Prasad V Potluri Siddhartha Institute of Technology Vijayawada520007 India Department of Electronics and Communication Engineering SRM TRP Engineering College Tiruchirappalli621105 India 

出 版 物：《Multimedia Tools and Applications》 (Multimedia Tools Appl)

年 卷 期：2025年第84卷第10期

页      面：6967-6997页

核心收录：

学科分类：0810[工学-信息与通信工程] 0830[工学-环境科学与工程（可授工学、理学、农学学位）] 080103[工学-流体力学] 08[工学] 0804[工学-仪器科学与技术] 0835[工学-软件工程] 0802[工学-机械工程] 0813[工学-建筑学] 0801[工学-力学（可授工学、理学学位）] 0812[工学-计算机科学与技术（可授工学、理学学位）] 

基　　金：No funding received by the authors 

主　　题：Temperature measurement 

摘      要：In academic institutions, the gathering of students within buildings significantly influences the ecological parameters of these spaces. Understanding and enhancing the environmental impact of academic buildings requires monitoring ecological parameters and crowd flow. The job of monitoring large-scale crowds and environmental conditions is a difficult one to do without difficulty. This study introduces reinforcement learning-based lightweight crowd flow measurement (RL-CFM) for real-time crowd flow tracking in institutional buildings. RL-CFM enables prompt responses to changes in crowd dynamics, enhancing its effectiveness during emergencies. The proposed RL-CFM periodically searches for smartphone-enabled requests, providing insights into crowd movement. Implemented and tested in an institutional building under real-world conditions, RL-CFM was installed in various locations, including the evacuation passage on the ground floor and two classrooms on the first floor. The study explores ecological parameters like temperature and CO2 concentration in the evacuation passage, considering various modes of smartphone operation that reflect people’s walking behavior. The RL-CFM’s performance is evaluated using different smartphone models with varying walking speeds, revealing a tracking accuracy of 94.32% in the Wi-Fi registered mode. © The Author(s), under exclusive licence to Springer Science+Business Media, LLC, part of Springer Nature 2024.