Augmented Reality on iOS for Guiding Novice Users Through Maintenance Tasks

Abstract

Written instructions and video tutorials work poorly for hands-on maintenance tasks: users split their attention between the instruction medium and the physical object. This paper presents an iOS application that uses ARKit to overlay step-by-step guidance directly onto the device or machine being serviced.

The system detects physical components using a Core ML model, anchors instruction overlays in 3D space, and advances through steps as the user completes each action. A user study compared the AR-guided approach against video instructions and printed manuals across a set of standardized maintenance tasks performed by participants with no prior experience.

The results show measurable improvements in task completion time and error rate. The paper also documents the platform-specific limitations of ARKit relevant to production deployment, particularly around tracking stability in variable lighting conditions.