Vision-Sensor Fusion-Based Low-Cost Dimension Measurement System for Machining Shop Floor

Abstract

Abstract The majority of industrial inspection activities across the globe involve considerable human efforts due to the requirements of robust detection abilities under varying lighting conditions, large field of view, higher speeds, and dynamic range of operation. In recent years, the motivation for integrating inspection and manufacturing operations on the shop floor has grown considerably to maximize the potential benefits of implementing Industry 4.0 initiatives. It is essential to digitize and integrate inspection of raw materials, in-process inventory, and finished components with manufacturing operations for achieving autonomous manufacturing goals. Machine vision offers an excellent alternative to manual inspection at a reduced cost, increased productivity, digital output, and negligible errors. However, its utility is restricted to the finished component inspection currently due to exorbitant equipment costs, requirements of specific lighting conditions, and a controlled operating environment. This paper presents the development of an integrated system combining a camera and ultrasonic sensor to develop a low-cost dimensional measurement system for inspection of raw material and in-process machined components in the shop floor setting. An indigenous inspection system is developed by integrating a camera, ultrasonic sensor, power source, and interactive user interface with onboard processing abilities. The augmentation of the ultrasonic sensor serves a dual purpose of eliminating the need for image calibration and dimensions measurement in the system. An interactive graphical user interface with acquisition trigger abilities and results display is developed using an open-source TkInter framework of the Python environment. A set of dimension measurements are performed for common raw material and in-process machined shapes using the developed setup. The comparison of results obtained using developed inspection setup and commercial machine vision system showed good agreement for common shapes. The efficacy of the developed system is validated for inspection of simple shapes with machining operations on the shop floor in the present work. The proposed system can be easily extended to inspect complex shapes such as gear profiles or camshafts as subsequent research work.