Quality monitoring of Concrete 3D printed elements using computer vision-based texture extraction technique

Concrete 3D Printing (3DP) has the potential to reduce construction time and the usage of labor and material in the construction industry. However, many parameters are found to influence the output of 3DP, and consequently, the variations in the quality of output are high. To fully realize the advantages of 3DP and to develop it into a technology for large-scale construction, quality monitoring and control are required. The workability of concrete is found to reduce with time, impacting the extrudability and buildability properties. This can be seen in 3DP elements where the bottom layers are found to have a smooth textural finish while the top layers have cracks, voids, and defects. To quantify the extrudability changes in the concrete, a new computer-vision-based methodology is proposed in this paper using a modified Histogram of Oriented Gradients (HOG) texture extraction method. Different levels of texture variations are extracted to quantify both minor and major textural changes. Weighted texture and normalized weighted texture metrics are introduced to have a combined single measure for minor and major textural variations. Further, a temporal textural change study is proposed to indirectly assess the buildability properties of concrete 3DP. This paper contributes to developing a non-intrusive autonomous quality monitoring and assessment technique for concrete 3D printed elements.