Deep learning technology for construction machinery and robotics

Ke You, Cheng Zhou, Lieyun Ding

Abstract

Construction machinery and robots are essential equipment for major infrastructure. The application of deep learning technology can improve the construction quality and alleviate the shortage of skilled workers, which is important for the future development of the construction industry. Deep learning has been recognized for its robustness in autonomous systems. A gap has been identified in the systematic analysis for deep learning-based technology for construction machinery and robotics applied in autonomous construction. To fill this gap, we conducted a systematic review from different perspectives, including: (1) perception; (2) navigation and planning; (3) control; and (4) human-robot interaction. On the basis of a systematic analysis, we identified the challenges applied to practice: (1) dataset limitation; (2) lack of interpretability; and (3) insufficient autonomous intelligence. Potential solutions and future outlook are as follows: (1) datasets with expert knowledge; (2) trustworthy artificial intelligence; (3) generative deep learning; and (4) extraterrestrial construction.

https://www.sciencedirect.com/science/article/abs/pii/S0926580523001127