Augmented vs. Traditional Instruction in Manufacturing Assembly
An Affordance-Based, Multi-Modal Assessment of Learning, Recall, and Retention
This site presents the v1.1 release of the dissertation. v1.1 corrects identified errata and render issues without adding new data or introducing new analyses. The exact changes are recorded in Appendix Appendix F — v1.1 Change Log. A PDF of the as-defended dissertation can be downloaded from the baseline release notes page. The project repository and source are public on GitHub.
A dissertation submitted to the Graduate Faculty of Auburn University in partial fulfillment of the requirements for the degree of Doctor of Philosophy
Approved Aug 3, 2024 by
- Dr. John Evans, Co-chair, Professor, Industrial & Systems Engineering
- Dr. Gregory Harris, Co-chair, Professor & Chair, Industrial & Systems Engineering
- Dr. Richard Sesek, Professor, Industrial & Systems Engineering
- Dr. Konstantinos Mykoniatis, Assistant Professor, Industrial & Systems Engineering
Abstract
This research rigorously assesses the effectiveness of augmented and mixed reality (AR/MR) instructional techniques in manufacturing training. By investigating how augmented methods designed to leverage specific affordances impact operator learning, recall, and retention in an authentic assembly context, it addresses critical gaps in understanding the relative efficacy of these technologies.
A cohort of 54 participants without prior exposure to similar tasks or AR/MR technology underwent simulated assembly training at the Tiger Motors Lean Education Lab. This between-groups study introduced participants to one of four instructional methods: Paper Work Instructions (PWI), Projected AR (PAR), head-mounted AR, and head-mounted MR. The same task was used across treatments, with minimal changes to instructional design. MR was differentiated from AR by allowing participants to freely manipulate the workpiece while maintaining aligned augmentations.
Performance metrics including task duration and error rates / types were carefully analyzed to assess learning progression and task proficiency. Measures of instructional reliance were also captured during recall, better informing the learning efficacy of each method. Several weeks later, retention was assessed on a volunteer subset to compare the durability of learning by treatment. The NASA Task Load Index and System Usability Scale were also administered to assess perceived workload and user experience. These data, coupled with demographics and qualitative feedback from open-ended exit interviews, contributes to a comprehensive expression of each participant’s experience.
Key findings revealed a nuanced trade-off between speed and accuracy across methods. Augmented technologies generally led to fewer errors but slower initial performance, while traditional methods allowed for faster execution but higher error rates. Recall and retention results were mixed, suggesting complex relationships between instruction method and long-term learning outcomes. Surprisingly, no significant differences in perceived workload were found across treatments, despite varying technological complexities.
This study’s implications extend beyond manufacturing to analogous tasks involving installation, repair, and safety training. Through its rigorous experimental design, ecological validity, and systematic operationalization of affordances, this research provides actionable insights to optimize AR/MR technology implementation and learning outcomes in industrial settings. The affordance-based framework introduced here offers a novel approach for evaluating and designing AR/MR training systems across various domains.