Using Augmented Reality and Step by Step Verification in Industrial Quality Control

Abstract

Quality control procedures are extremely important among industrial applications. Generally, these tasks include many repetitive tasks that require manual intervention. Given their complexity, quality control tests are often detailed in video recordings, paper instructions, photos or diagrams to guide workers throughout the process. Augmented Reality (AR) has been making significant progress in the last decades, becoming mature enough to be used in industrial scenarios. While some AR systems have been proposed to support quality control procedures, most of them only present information to workers but do not track or validate the process in real-time being used only to guide it. Another limitation of existing systems is the generation of virtual instructions used by AR systems to guide the operator. In this work, we propose an AR-based tool to guide users by overlaying information in a video stream while performing real-time validation during the execution of quality control procedures. The main objective is to provide dynamic support and decrease the mental workload needed to complete the procedure as well as the number of errors, facilitating the procedure execution by untrained workers. Besides this, the tool allows to create virtual content that can be used to generate step-by-step instructions automatically based on human demonstrations. By making the virtual instruction creation effortlessly it is possible to eliminate the user’s need for memorizing new instructions with each change of the product lines. While presenting task relevant information the system uses computer vision techniques to keep track of the procedure stage, verifying its completion and switching automatically to the next step without requiring any interaction from the user. A comparison between the time taken to perform the procedure with and without validation was made. The results show that the validation process would confer the process a significant efficiency boost, while avoiding possible human errors.