Augmented Reality (AR) holds significant potential to facilitate users in executing manual tasks. For effective support, however, we need to understand how showing movement instructions in AR affects how well people can follow those movements in real life. In this paper, we examine the degree to which users can synchronize the speed of their movements with speed cues presented through an AR environment. Specifically, we investigate the effects of timing in AR visual guidance. We assess performance using a highly realistic Mixed Reality (MR) welding simulation. Welding is a task that requires very precise timing and control over hand and arm motion. Our results show that upfront visual guidance (before manual task execution) alone often fails to transfer the knowledge of intended speeds, especially at higher target speeds. Live guidance (during manual task execution) during the activity provides more accurate speed results but typically requires a higher overshoot at the start. Optimal outcomes occur when visual guidance appears upfront and continues during the activity for users to follow through.

Cleanroom cleaning is a surface coverage task where the pattern should be followed correctly, and the entire surface should be covered. We investigate the efficacy of augmented reality (AR) by implementing various pattern guidance designs to enhance a cleanroom cleaning task. We developed an AR guidance system for cleaning procedures and evaluated four distinct pattern guidance methods: (1) breadcrumbs, (2) examples, (3) middle lines, and (4) outlines. We vary the instructions on the entire surface or as a single step. To measure performance, accuracy, and user satisfaction associated with each guidance method, we conducted a large-scale (n=864) between-subjects study. Our findings indicate that single step instructions proved to be more intuitive and efficient than full instructions, especially for the breadcrumbs. We also discussed the implications of our results for the development of AR applications for surface coverage and pattern optimization.

Text input in Virtual Reality (VR) is crucial for communication, search, and productivity. We compared four keyboard designs for VR text entry, leveraging the flexibility and the tracking options of a 3D environment. We used the Dvorak layout to control for experience differences. The designs were: (a) a floating keyboard with touch input, (b) a keyboard attached on the back of the hand with touch input, (c) a floating keyboard with eye tracking and pinch input, and (d) a keyboard laid out over a rolling shape with touch input. Designs (b), (c), and (d) can move in 3D space, while design (a) is static. Design (d) had similar efficiency to design (a) but with better usability and lower Physical Demand. Design (b) led to higher Physical Demand, Effort, and Frustration. Design (c) had lower Physical Demand but higher Mental Demand, Effort, and error rates. Typing speeds averaged 6.51 WPM (1.24% error rate) for (a), 5.56 WPM (3.82% error rate) for (b), 5.33 WPM (1.43% error rate) for (c), and 6.70 WPM (1.64% error rate) for (d).

In many jobs, workers execute precise line tracing tasks; welding, spray painting, or chiseling, for example. Training and support for such tasks can be done using VR and AR. However, to enable workers to achieve the required precision in movement and timing, the effect of visual guidance on continuous movement needs to be explored. In VR environments, we want to ensure people are trained so that the obtained skill is transferable to a real-world context, whereas, in AR, we want to ensure an ongoing task can be completed successfully when adding visual guidance. To simulate these various contexts, we employ a VR environment to investigate the effectiveness of different visualizations for motion-based guidance in a line tracing task. We tested five different visualizations, including faster and slower arrows on the pen, the same arrows on the line, a dynamic graph on the pen or line, and a ghost object to follow. Each visualization was tested with the same set of five lines of different target speeds (2cm/s to 10 cm/s in steps of 2 cm/s) with a training line of 5 cm/s. Our results show that the example ghost on the line turns out to be the most efficient visualization for allowing users to achieve a specific speed. Users also perceived this visualization as the most engaging and easy to use. These findings have significant implications for the development of AR-based guidance systems, specifically in the realm of speed control, across diverse domains such as industrial applications, training, and entertainment.