Evaluation of the effectiveness of robotization of product quality control operations using industrial manipulators and visual sensor systems

Abstract

This paper reveals the transformation of the processes of ensuring the quality of construction products through the integration of industrial manipulators with visual sensor systems, which radically changes the paradigm of acceptance control of reinforced concrete structures, metal products and facade systems in conditions of high variability of typesizes and complex spatial geometry. Adaptive scanning trajectories are considered using six-axis KUKA, Fanuc, and ABB robots equipped with structured light cameras, laser profilometers, and stereo systems that generate point clouds with a density of up to several million measurements per product and register them relative to digital references using iterative nearest neighbor methods using the Open3D and PCL libraries. Quantitative analysis demonstrates a reduction in the control time of a standard panel from 18-22 minutes for manual measurement to 94 seconds for automated measurement, while simultaneously increasing the resolution and reducing the error to 0.018-0.067 mm, depending on the configuration, including hybrid schemes that optimally balance performance and accuracy in the presence of abrasive dust, temperature fluctuations, and variable illumination. Machine learning algorithms allow you to achieve an F1-measure of over 0.9 when recognizing cracks, chips, shells, and spills, with significant environmental parameter identification through correlation and regression analysis and recommendations for maintaining operating ranges. Economic modeling based on long-term monitoring data of large enterprises confirms the payback period of capital investments in 2.7-3.0 years, positive values of NPV and IRR above 31%, due to the release of personnel, a reduction in hidden defects by 3-4 times, and an increase in the productivity of control posts by more than four times. A multivariate regression model with R² = 0.847 establishes the determining role of output volume and nomenclature stability in integral efficiency, indicating the need for parameterized models for small-scale production. The directions of regulatory support, integration of results with BIM technologies, creation of digital product passports and prospects of self-learning systems are discussed, which lays the foundation for a new technological order in the construction industry, increasing the objectivity, traceability and reliability of building structures throughout the entire life cycle. These provisions allow to assess the relevance of the article for specialists in the field of robotics, metrology and digital transformation of production.