Particulate matter source localization in dynamic indoor environments: Bridging simulation-experimentation gaps with a 3D multi-robot system

Indoor particulate matter (PM) threatens human health, compromises product quality and yield, and poses safety risks. Dynamic indoor environments introduce unpredictable airflow changes and complex PM behavior, posing significant challenges for accurate PM source localization by mobile robots. This study introduces our self-developed multi-robot system for three-dimensional (3D) concentration detection, bridging the gap between simulation and real-world applications. A total of 225 experiments across 15 cases were conducted in a dynamic ventilation environment, enabling a comprehensive analysis of airflow dynamics and PM behavior. The improved whale optimization algorithm (IWOA) was compared in 3D and 2D scenarios and further evaluated against the improved particle swarm optimization (IPSO) method under 3D conditions. The adaptability of IWOA_3D to variations in PM size, release rate, source location, and accuracy standard was also investigated. Using the success rate as a key evaluation criterion, IWOA_3D demonstrated strong adaptability to variations in source height, localizing PM2.5 at 1.05 m with a 73.3% success rate, despite limitations related to PM size and source location. These findings highlight the importance of selecting appropriate PM sensor readings and optimizing localization strategies in dynamic indoor environments, while demonstrating the practical effectiveness of the IWOA_3D method.