Augmented Reality Based Object Detection and Rendering for Industrial Machines using Visual SLAM Based Deep Learning

Abstract

The goal of Industry 4.0, which is at the vanguard of the current industrial revolution, is to efficiently meet the varied needs of stakeholders by integrating digital and physical domains into industrial processes. This research details an innovative Pervasive Augmented Reality (AR) prototype developed in tandem with renowned business partners. Making it easier to detect problems in an industrial assembly line and keep tabs on data in real time is the main goal of this prototype. The major focus is on the preprocessing steps, semantic segmentation, and object recognition that make the model usable as a scene for different kinds of maintenance tasks, such scheduling maintenance away from the site or redesigning it. We used a human-centered design (HCD) approach to learn about the challenges, needs, and complexities of the stakeholders, and to provide the groundwork for how the model should develop. These people’s first impressions are outlined below, after an initial user research in a manufacturing shop floor environment to evaluate the prototype and get feedback on how to make it work better under such conditions.

Keywords: Augmented reality, Deep learning, Human centered design, Industry 4.0, User interface.

View PDF