Quantifying Movement Behaviors in VR
In recent years, virtual reality (VR) has shown promise as a rehabilitation tool to restore arm function in stroke patients. At the same time, researchers are more frequently using kinematic metrics to convert motion tracking data into useful insights about patients’ progress.
Since modern consumer VR systems already collect the data needed to calculate many common kinematic metrics, these systems could eventually be used to both deliver stroke rehabilitation programs and administer kinematic assessments to monitor patients’ recovery. However, it is not yet clear how the properties of VR-based assessment tasks may influence kinematic metrics that are used to assess arm function post-stroke.
To begin addressing this question, we examined the influence of two task properties (movement direction and hand dominance) on a set of 10 kinematic metrics during a discrete reaching task performed by healthy participants using an Oculus Quest 2 VR headset. Our results provided an initial account of how these kinematic metrics were influenced by each task property. These findings were published as late-breaking work at CHI 2022.
A previous study using traditional kinematic analysis techniques yielded some unexpected results, calling into question the extent to which the strategies users employ when moving to physical targets generalize to virtual reality environments. Although we proposed several potential explanations for these unexpected results, we were unable to discriminate between these explanations using traditional one dimensional kinematic analyses.
To explore the source of our unexpected results, we re-analyzed the data from our most recent study using 3D analysis techniques borrowed from the movement variability literature. The results of this analysis shed light on the potential source of the unexpected results and highlight the importance of considering both physical and perceptual influences on movement strategies when analyzing movement behaviors in VR.
Kinematic analysis of people’s behavior during goal-directed reaching movements has revealed that we unconsciously adopt different types of strategies to maximize performance and minimize the amount of energy we need to expend to complete our movements. Thanks in large part to the work of Digby Elliott and colleagues, we know a lot about the strategies people use to select physical targets. However, it is not yet clear if or how people might adjust their strategies in virtual reality (VR) environments.
To examine this question, we adapted a task from previous laboratory studies to examine strategic movement biases in VR. Participants selected targets at different locations in a VR interface, and we recorded and parsed their movement trajectories to highlight informative milestones in each target selection movement. Our results showed that participants took a similar amount of time to complete the task across all conditions, but they used different strategies to select targets at different locations. These findings confirm that people can adapt their movement behaviors to optimize movement performance in VR. Our results also highlight how people adjust their movement strategies to cope with new challenges introduced by VR environments.
As virtual reality (VR) interfaces become more widespread, specialized metrics may be needed to assess the efficiency of complex 3D interactions with these interfaces. To address this need, we explored whether kinematic movement trajectory analysis, an established technique in experimental psychology, could provide a useful framework for quantifying meaningful patterns of user interaction using the 3D position data recorded by most VR displays.
Participants selected target objects of varying perceptual ambiguity presented in a simple VR environment with limited depth cues. We used a-priori criteria to examine movement trajectories during this task and identify discontinuities consistent with movements toward incorrect target locations (“misfires”). We found that movements with the kinematic properties expected of misfires occurred significantly more often during movements to perceptually ambiguous targets. These results suggest that kinematic measures may be a useful tool for quantifying patterns of user interaction in VR interfaces.
Virtual reality interfaces often require users to perform three-dimensional reaching and pointing movements to interact with objects positioned within arm’s reach. However, there has been limited work to evaluate the applicability of established models of human motor control for predicting performance on these tasks in 3D virtual environments. In this study, participants performed a 3D discrete pointing task in a virtual environment presented using an Oculus Rift system. The data from this task were used to identify factors that influence movement performance in VR and develop a new formulation of Fitts’ law that accounts for these influences.
Extended models that account for the effects of target depth yielded better predictive power than the traditional Fitts’ law formulations used to model 2D hand and mouse movements. This result points to a need for extended Fitts’ law models to account for the unique constraints imposed by VR environments.
