Exploring Localization Methods for Monitoring Indoor Construction Using Support Vector Machines (SVMs) and Deep Neural Networks (DNNs)

Document Type

Conference Proceeding

Publication Title

Lecture Notes in Civil Engineering

Publication Date

1-1-2026

Abstract

Monitoring construction workers on-site generates critical data to enhance safety and productivity, especially in high-risk and dynamic construction environments. Traditional tracking methods like GPS and computer vision-based systems are commonly employed in outdoor settings, but these approaches are often limited by signal strength and line-of-sight requirements, making them less effective indoors. Indoor construction sites pose additional challenges, as the constantly changing environment—due to moving equipment, metallic structures, and physical obstacles—frequently disrupts signal propagation and reduces the accuracy of conventional localization techniques. To address these challenges, this study investigates the use of IoT technology and machine learning to overcome the limitations of existing tracking methods and to adapt to the complexities of indoor construction settings. Specifically, the research utilizes a newly introduced Bluetooth Low Energy (BLE) protocol called Angle of Arrival (AoA), which enhances the precision of indoor localization. A six-phase experimental approach was conducted within a three-story laboratory building on a university campus, with each floor outfitted with strategically positioned sensors and locators to track worker trajectories in real-time. The data gathered from these experiments was processed using machine learning algorithms to filter environmental noise and dynamically adjust to fluctuating site conditions. Experimental results showed a marked improvement in localization accuracy and robustness over the six phases, achieving a final accuracy of over 90%. This level of precision demonstrates the viability of the proposed method for real-world applications in indoor construction environments. In summary, the proposed approach offers a scalable solution highlighting the potential of BLE AoA-based localization combined with machine learning to transform construction safety protocols and project coordination, effectively meeting the industry's evolving needs.

Volume

823 LNCE

First Page

625

Last Page

638

DOI

10.1007/978-3-032-18712-3_44

ISSN

23662557

E-ISSN

23662565

ISBN

[9783032187116]

Link to Full Text

Share

 
COinS