Start Colloquium Kahao Cheung

Reinforcement Learning for 3D Concrete Printing

Within the built environment, the digitization of designs, processes and manufacturing methods has experienced a significant growth during the last decades. One of the emerged manufacturing methods is 3D concrete printing (3DCP) which uses a layer-wise printing strategy to construct building elements such as walls, columns and other construction elements.
Although the prospects of successful 3DCP are highly beneficial to the construction industry, challenges such as its current trial-and-error-based approach still prevent its widespread use.

To transition from a trial-and-error methodology towards first-time-right manufacturing, it is essential to control the variations in the 3D concrete printing process.

With the rapid increase in machine learning (ML) implementations in the construction industry and in 3D printing applications, new opportunities arise from data-driven strategies for efficient and accurate prediction and control.
In this research, a proposal will be made for a framework that uses machine learning to anticipate for variations in the 3DCP process.