Use of optical systems in assembly process and quality control of automotive engines

The goal of this thesis project is to describe how the use of industrial vision systems can improve the productive and commercial results of a company producing internal combustion engines for automotive application. The study has been conducted in an automotive powertrain plant, belonging to an important car manufacturing group operating worldwide. With the support of the production engineering office it was possible to follow the activity of the various optical systems present along the machining and assembly lines of the plant, being on the field. The introductive part of this dissertation focuses on a general overview of the industrial vision field, pointing out its main advantages and illustrating its leading applications. Furthermore, for a better understanding of the logical flows connecting the different operations and how vision solutions interact with the product in the process, it is reported an analysis of the three engine models produced and of their assembly processes. The most advanced vision systems implemented in the plant of study fall in five different categories: traceability, silicone adhesive path application, metrology, assembly quality check and final product quality inspection. Concerning the traceability issue, the analysis has been carried out considering the solutions relative to barcodes, data matrix codes and engine serial numbers. This class of systems have been compared to the Radio-Frequency Identification (RFID) one, which is present in the plant too. Optical control systems dedicated to the silicon adhesive path application are necessary to verify that the sealing injected bead has a specific width and height, following correctly the profile indicated. Timing cover and oil sump installation require this kind of control. Besides the presence on the line, vision systems are adopted also for metrology purposes both in the metrology laboratories in which measuring devices and reference components are inspected, and on the gauging stations located nearby the production lines where pieces coming from different batches are systematically analyzed. The majority of the optical systems are operated to control the presence and orientation of elements like screws and caps. A far advanced technology included in this category is the 3D control of the orientation of piston-pin circlip. Finally, when the engine is complete it passes through a machine vision system composed of multiple robotic arms endowed with cameras taking pictures of the engine from different angles and comparing them with a set of reference pictures. In this way, defects that could stay hidden until the last stages of the process are detected. In the conclusion of this thesis work possible future trends of industrial vision systems are discussed, also reporting the growing importance of the deep learning technique.