A Computer Vision and Point Cloud-Based Monitoring Approach for Automated Construction Tasks Using Full-Scale Robotized Mobile Cranes

Recent years have witnessed rapid development and contemporary trends in smart construction research owing to advances in machine learning algorithms, modern sensory systems, and robotic technologies. In this paper, a novel economical computer vision and point cloud-based monitoring framework is proposed to assist in the lifting and relocation of construction sources via mobile cranes on site. The proposed framework incorporates a multicamera approach to achieve multiple goals, such as 3D vision-based real-time reconstruction, 3D localization of construction resources, and safety monitoring. To demonstrate the effectiveness of the proposed framework, field experiments were conducted on a full-scale mobile crane. The results show that the proposed monitoring system achieves real-time performance, which can successfully recognize construction resources and guide the crane to initialize the lifting position and avoid potential moving workers during motion execution.