John Karasinski

Document Type



Despite the extensive previous research into manual tracking, to the author’s knowledge there exists no study in the literature addressing human performance or workload changes in manual tracking tasks between traditional computer monitors and mobile, augmented reality headsets. The aim of this study was to investigate the effect of several factors on human performance and workload in a three-axis manual tracking task. Recent advances in computing hardware have enabled a new generation of augmented reality stereoscopic devices, such as the Microsoft HoloLens, which have yet to be evaluated in the literature. The Microsoft HoloLens provides depth cues to the user which are not available with traditional 2D displays. If 3D displays can successfully improve performance and decrease workload in a manual tracking task, then it may also be valuable in the realm of robotics tasks. The human control of spacecraft, underwater, surface, and surgical robotics operations are all exciting areas of potential benefit. Each of these areas poses unique domain specific challenges, but the role of the human controller in robotic operations is often based on a simple three-axis tracking task. We designed an experiment that investigated if the depth cue offered by 3D displays could improve performance and decrease workload. With the knowledge that We were also interested to see if a 2D display could gain the benefits of a depth cue by rotating the axis of the task such that the depth cue was more readily available. Research has shown that presenting three-dimensional information on a two-dimensional screen is not a simple task, and that the projection of the 3D information onto the 2D screen can cause large changes in the performance of the user [3]. In addition to these cues, we also investigated the effects of concurrent bandwidth feedback on task performance and workload as an alternate technique to improving performance. Concurrent bandwidth feedback alerts the operator when their real-time performance has drifted outside an acceptable, predefined window of performance. The use of feedback has been shown to improve performance in a wide variety of motor control tasks [4, 5]. We added this countermeasure as we have previously found it to be effective in similar tasks [6]. This study assessed the influence of display type (perspective vs. stereoscopic), relative display attitude (zero degrees vs. thirty degrees), and concurrent bandwidth feedback (with vs. without) on performance and workload. Objective performance was measured using the root-mean-square error (RMSE) of the depth (z) axis, and subjective performance was measured with the use of a questionnaire. Objective workload was measured using the response time to a secondary, two-choice task, and subjective workload was measured using the NASA-TLX [7]. It was hypothesized that: 1. Concurrent bandwidth feedback improves performance in the depth (z) axis for both display types and will decrease workload. 2. Stereoscopic augmented reality displays improve performance in the depth (z) axis, but do not affect workload. 3. Rotating the display improves performance in the depth (z) axis for both display types and will decrease workload. A human-in-the-loop simulation was conducted using a fixed-base simulator, see Figure 1. The simulator consisted of two 10.4-inch LCD displays. The tracking task was completed on the left display, while the right display showed the two-choice task. For participants in the 3D group, the left LCD monitor was turned off, and the tracking task was instead displayed on the HoloLens. Subjects in both groups used the same Microsoft Xbox controller and control scheme to complete the task. Before entering the study, participants were randomly placed in a display group, and were then randomly placed into an order group (which consisted of Baseline-Feedback-Rotated, Feedback-Rotated-Baseline, and Rotated-Baseline-Feedback). This order group was created to remove any order effects that might arise due to training on a given display. The order of the designs was expected to be insignificant, but we will later discuss how this is not the case. Three designs were presented to the subjects to evaluate: a baseline design, a color-based concurrent bandwidth feedback (CBF) design, and a rotated design. Figure 2 shows all three designs in the same error state. The three designs were very similar, having only minor differences between each other. The baseline design consists of a flat cross with a center target point and a green sphere error indicator. This indicator also casts a green, variable-length rod perpendicular to the plane of the cross, which allows for a visual estimation of the error in the z-axis. The x-axis is parallel with the horizontal cross, while the y-axis is parallel with the vertical cross. The color feedback display was identical to the baseline design in every way, with the additional of visual concurrent bandwidth feedback on the z-axis. When the absolute value of the error on the z-axis exceeded a fixed bandwidth, the color of both the spherical indicator and the cylindrical rod changed from green to red. When the absolute value of the error on the z-axis was lowered back below this fixed bandwidth, the indicator changed back to a green color. The rotated display was identical to the baseline design, but the relative attitude of the display was rotated about the y-axis by 30 degrees.

Publication Date



Link Foundation Fellowship for the years 2017-2018.

FORM Final Report John Karasinski.pdf (93 kB)
Standard cover form for report



To view the content in your browser, please download Adobe Reader or, alternately,
you may Download the file to your hard drive.

NOTE: The latest versions of Adobe Reader do not support viewing PDF files within Firefox on Mac OS and if you are using a modern (Intel) Mac, there is no official plugin for viewing PDF files within the browser window.