Abstract—This paper presents research on development of a novel automatic quality inspection station for concrete products using a cobot equipped with a force sensing finger and embedded machine learning model. After a concrete product is made, it is cured for 28 days to gain full strength. Then, its quality is assessed. However, it is highly desirable to quickly and accurately assess the product quality moments after it comes out of the mold as so called “green”, uncured, no-slump concrete to eliminate waste and improve quality. Currently, a human operator inspects the green concrete products by poking and visual inspection as they come out of the molds. This is a highly subjective and often inaccurate approach. Experimental results with the cobot showed 92.3% accuracy in predicting quality of concrete blocks compared to the human accuracy of 50%. The new inspection system can be a viable solution to predict quality of resulting cured concrete blocks from initial tests of green concrete products during production. The system can alert for production problems early on leading to reduced costs and increased product quality when cured.

Index Terms—Cobot, concrete, mechatronics, robotics, machine learning, UR10

Cite: Aaron Burke and Hakan Gurocak, "Automatic Inspection of Green Concrete Quality Using Machine Learning and Cobot," International Journal of Mechanical Engineering and Robotics Research, Vol. 11, No. 5, pp. 331-337, May 2022. DOI: 10.18178/ijmerr.11.5.331-337

Copyright © 2022 by the authors. This is an open access article distributed under the Creative Commons Attribution License (CC BY-NC-ND 4.0), which permits use, distribution and reproduction in any medium, provided that the article is properly cited, the use is non-commercial and no modifications or adaptations are made.