AR on Ladders: Effect of Performing Augmented Reality Tasks on Postural Sway While Working on a Ladder

The use of augmented reality (AR) is growing in industrial occupational settings such as within the construction industry [1, 2]. While AR provides convenience such as providing hands-free visual access to time-sensitive information, it may also introduce unintended adverse consequences. Falls off ladders are already a major source of workplace injuries [3], and any adverse effects of AR on balance while working on a ladder could exacerbate the risk of falling. Therefore, the purpose of this study was to investigate the effects of viewing and interacting with AR on balance during simulated construction tasks while on a ladder. The four tasks investigated involved: (1) a reading task with AR input, (2) an assembly task requiring AR input, (3), an assembly task with verbal input, and (4) a button poking task with virtual AR buttons. We also investigated screen-relative and world-relative AR presentation styles across the relevant AR-support tasks. Balance was assessed by measuring mean speed of postural sway of the center of pressure using a force plate under the ladder, with increases in sway speed from a baseline condition involving standing as still as possible on the ladder being compared between tasks and presentation styles. Results indicated postural sway was greater during the assembly task with verbal input (6.39 cm/s; p<0.05), the assembly task with AR input (6.33 cm/s; p<0.05), and the button poking task with AR input (7.35 cm/s; p<0.05) as compared to the reading task wi th AR input (2.90 cm/s). Additionally, postural sway increased more during the button poking task with AR input as compared to the assembly task with AR input. Regarding the two presentation styles, postural sway with a world-relative AR display (9.27 cm/s) was greater as compared to a screen-relative AR display (4.95 cm/s) during the poking task with AR input (p < 0.05) but did not differ between presentation styles during the assembly task with AR input (p > 0.05) or the reading task with AR input (p > 0.05). Our results imply that care must be taken when designing AR applications for applications that are to be used on ladders or other environments where postural stability is of great importance, as the task to be completed and the presentation style can affect a worker's balance. Further research into user interface presentation styles interaction techniques and tasks involving AR menu navigation is proposed.