Assessing AR Visualizations during Non-Complex Assembly: Reporting Preliminary Results from a Crowdsourcing Rework Study

This study assesses the viability of using a simple assembly task as part of a crowdsourcing experiment to evaluate different visualizations in augmented reality (AR). We recruited 36 participants through a crowdsourcing platform to assess how visualization types (i.e., none, contour, and abstract) impact the performance of a gear-box assembly rework task. Our preliminary results suggest that the learning effect associated with manual assembly of a non-complex gearbox object plays a more significant role in improving performance than the type of visualization used. This pilot study also highlights requirements for crowdsourced task difficulty in order to offer potentially viable insights.