Component-based BIM-semantic web integration for enhanced robotic visual perception

Visual perception plays a crucial role in enhancing construction robots’ real-time performance by enabling accurate object recognition, localization, and scene understanding in complex construction environments. While current research focuses on visual recognition for tasks like spatial localization and navigation, it falls short of addressing the more advanced functions necessary for comprehensive construction planning. This paper proposes an integrated framework that combines building semantic web technologies with image recognition techniques to significantly improve robotic perception. By merging real-time image recognition with BIM-based semantic mapping, robots can gather critical information on component types, spatial relationships, and construction requirements. Case studies illustrate the framework’s ability to improve adaptability, precision, and efficiency in both static and dynamic construction environments, thereby enabling more intelligent, automated, and efficient robotic construction processes, with the potential for broader applications in the future of construction technology.