<p>The traditional manufacturing system involves a separate inspection of the roundness error of the machined component which increases the overall manufacturing time and cost of the component. The accuracy of the roundness error in the traditional roundness measuring system is affected by the component’s true centering. A system for on-machine roundness measurement using machine vision was developed to overcome such issues. During the measurement process, a novel approach was used in which, on the same image, calibration and measurement were performed. In this study, a novel Cyclic-ANN methodology was proposed to predict the diameter and roundness errors of the component. The predicted value of the roundness error using Cyclic-ANN methodology was found to be 0.0142 mm which was compared with the CMM measurement of the component with a value of 0.0154 mm. Thus, experimental results show the robustness, accuracy, and simplicity of the developed system. The proposed methodology estimates the ‘actual working roundness error’, allowing the design engineer to modify the component’s design tolerance and the manufacturing engineer to modify the machining operations to reduce manufacturing costs.</p>