Search

Head Mounted Display (HMD) based Augmented Reality (AR) is being increasingly used in manufacturing and maintenance. However, limited research has been done to understand user interaction with AR interfaces, which may lead to poor usability, risk of occupational hazards, and low acceptance of AR systems. This paper uses a theoretically-driven approach to interaction design to investigate the impact of different AR modalities in terms of information mode (i.e. video vs. 3D animation) and interaction modality (i.e. hand-gesture vs. voice command) on user performance, workload, eye gaze behaviours, and usability during a maintenance assembly task. The results show that different information modes have distinct impacts compared to paper-based maintenance, in particular, 3D animation led to a 14% improvement over the video instructions in task completion time. Moreover, insights from eye gaze behaviours such as number of fixations and transition between Areas of Interest (AOIs) revealed the differences in attention switching and task comprehension difficulty with the choice of AR modalities. While, subjective user perceptions highlight some ergonomic issues such as misguidance and overreliance, which must be considered and addressed from the joint cognitive systems' (JCSs) perspective and in line with the predictions derived from the Multiple Resources Model. [ABSTRACT FROM AUTHOR]

In Industry 4.0, integrated data management is an important challenge due to heterogeneity and the lack of structure of numerous existing data sources. A relevant research gap involves human knowledge integration, especially in maintenance operations. Augmented Reality (AR) can bridge this gap, but it requires improved augmented content to enable effective and efficient knowledge capture. This paper proposes dynamic authoring and hybrid recommender methods for accurate AR-based reporting. These methods aim to provide maintainers with augmented data input formats and recommended datasets for enhancing the efficiency and effectiveness of their reporting tasks. The proposed contributions have been validated through experiments and surveys in two failure diagnosis reporting scenarios. Experimental results indicated that the proposed reporting solution can reduce reporting errors by 50% and reporting time by 20% compared to alternative recommender and AR tools. Besides, survey results suggested that testers perceived the proposed reporting solution as more effective and satisfactory for reporting tasks than alternative tools. Thus, proving that the proposed methods can improve the effectiveness and efficiency of diagnosis reporting applications. Finally, this paper proposes future works towards a framework for automatic adaptive authoring in AR knowledge transfer and capture applications for human knowledge integration in the context of Industry 4.0. [ABSTRACT FROM AUTHOR]