Implementation of AI for Injector Quality Inspection

Abstract

A fuel injector is an integral part of the fuel injection system. It is a device that actively injects fuel into an internal-combustion (IC) engine by directly forcing the liquid fuel into the combustion chamber at an appropriate point in the piston cycle. Quality inspection of an assembled injector includes checking the injector for dents, rust, cracks, O-ring coloring and cuts, correct assembly, filter screen fitment and damage and cup inspection to name a few. Until now the injector inspection has been performed visually by a human operator. Skilled as they may be , human errors are bound to happen. What may seem as a small and insignificant defect, may cause functioning issues for the engine in the future. This project aims to convert the physical inspection into an automated inspection process seamlessly, to reduce the operator dependency and to ensure 100% quality, by implementing an autonomous and AI driven system with computer vision that will check the injector for aesthetic defects and will give the final injector OK/NOT OK decision, while also ensuring the correct assembly. Thus, transformation of business processes by implementing new digital technologies and automation will result in opportunities for efficiency and will increase revenue.