A Review on Augmented Reality Tracking Methods for Maintenance of Robots

Abstract

Augmented reality (AR) in maintenance is a broad subject with many nuances when it comes to their implementation. The applications of these systems range from maintenance of large-scale assets such as buildings to smaller scale assets such as robots. Applications of AR in maintenance typically serves as a visual guide to assist users in diagnosis or steps needed to be performed for maintenance. In this paper, the tracking methods utilized in AR-based maintenance for robots are qualitatively evaluated. The reviewed works in this paper are between the years of 2015 to 2020 to ensure that the AR tracking methods are relatively state of the art. It is found that applications of AR-based maintenance for robots are uncommon in the scope defined in this research especially in the industrial environment as most reviewed works are conducted in a laboratory setting. In addition to that, it is found that marker-based tracking methods are commonly utilized in these applications.

Advanced Visual SLAM and Image Segmentation Techniques for Augmented Reality

“…reviewed the most recent deep learningbased tracking methods and divided them into three categories based on network structure, functionality, and training. The tracking methods used in AR-based robot maintenance are qualitatively evaluated (Koh et al, 2020). Jiao et al (2021) reviewed the critical advances made by deep learning, including deep feature representations, network architecture, and four critical issues in visual tracking (e.g., spatiotemporal information integration, target-specific classification, target information update, and bounding box estimation).…”