Open Collections

Xiao, Yifei

University of British Columbia

With the rapid population growth worldwide, a large amount of new infrastructure needs to be constructed. However, the rapid demand for new construction is challenged by lack of skilled workers, long construction time, poor quality assurance and weather issues. The adoption of new technologies in the current construction industry is considered slow, resulting in a stagnation in its productivity. To tackle these challenges, one possible solution is to use robots and artificial intelligence (AI) algorithms for construction. Compared to the current mainstream and widely studied robots (such as unmanned aerial vehicles, unmanned aerial vehicles, and robotic arms), robotic cranes would be more suitable for heavy construction tasks due to their large payload and wide working range. However, few robotic cranes are currently used for construction practice and AI algorithms have not been widely applied in the construction industry. The backwardness of current construction technology has prevented the construction industry from transforming towards digitalization and intelligence to improve productivity and meet current and future demand challenges. To address these limitations, an AI-based hierarchical framework is proposed to robotize a traditional mobile crane for construction practice. This framework includes high-level AI algorithms and low-level robot control algorithms. The high-level AI algorithms are responsible for localizing construction objects and obstacles on the construction site, and then developing a collision-free lifting strategy to transport the identified construction object to a target position. This is achieved using deep learning and reinforcement learning algorithms, respectively. On the other hand, the low-level algorithms are responsible for regulating motions of a robotized crane using robotic kinematics and advanced nonlinear backstepping control methods. Furthermore, to achieve real-time construction process monitoring, novel hardware communication interfaces are developed between building information modelling (BIM) platform, robot operating system (ROS), and the robotized crane. To validate the proposed framework, simulation tests and full-scale retaining wall construction experiments have been conducted. Results show that the robotized crane can automatically construct the retaining wall with collision avoidance by using the proposed AI-based hierarchical framework. The entire construction process can be monitored in real time using the developed hardware communication interfaces.

Text

2024-12-10

Attribution-NonCommercial-NoDerivatives 4.0 International

http://hdl.handle.net/2429/89817

Civil Engineering

University of British Columbia

UBCV

http://creativecommons.org/licenses/by-nc-nd/4.0/