Publications / 2024 Proceedings of the 41st ISARC, Lille, France
Construction progress monitoring is a crucial aspect of project management, ensuring project completion on time, within budget, and meeting quality standards. Current data collection and processing methods are time-consuming and labor-intensive, and modern technologies, including robots and computer vision, offer a potential solution to this challenge. While past research has improved data collection and processing independently, a significant manual effort is still required to classify and transform collected data for processing. This research introduces the foundation of a novel framework to bridge this gap, leveraging an autonomous four-legged robot using the Software Development Kit (SDK) and the Robot Operation System (ROS). Additionally, Simultaneous Localization and Mapping (SLAM) for navigating, and cloud computing platforms for data processing, are employed to establish a repository for raw data from images captured by the robot, facilitating simultaneous processing stages. The research proposes a cloud-based machine-learning service that hosts a deep-learning algorithm for material identification and quantification. The objective of the proposed framework and system architecture is to capture images of the indoor environment of construction projects, classify and process data to create quantified material lists, and compare them to the full material list on the cloud, construction schedules can potentially be updated in real-time.