A Comparison between Virtual Reality and Augmented Reality on Upper-limb Prosthesis Control

Abstract

In recent years, virtual reality (VR) and augmented reality (AR) technologies have been shown to be promising avenues for improving the security, convenience, and efficacy of rehabilitative prosthesis training systems. Despite their rise in popularity, it is still unclear what advantages these paradigms have over one another when applied to complex motor tasks. In this study, we aim to determine which paradigm, AR or VR, is better suited for the completion of dexterous motor control tasks needed for effective upper-limb prosthesis use. We evaluate a population of able-bodied (N=5) subjects. Each of them performed 50 3-dimensional object manipulation tasks in analogous AR and VR environments respectively, with 100 trials for each subject. During each trial, subjects operate a virtual upper-limb prosthesis to perform reach-grasp-relocation manipulations via a myoelectric pattern recognition (MPR) algorithm. We report an average improvement in Fitts’ throughput (+20.94% and +21.26%) from all subjects when comparing VR to AR task performance in the reach and relocation phase. Additionally, we observe an increase in overall task completion rate (+3.60%) and mean path efficiency (+9.59% and +6.73%) during the reach and relocation phases of motion. What's more, we report a statistically significant decrease in mean task completion time during both reach and relocation phases when comparing AR to VR-based trials (p<0.05). Based on these functional results, we conclude that as a paradigm, VR promotes more efficient motion, resulting in higher task completion rates and path efficiency. On the other hand, AR allows subjects to perform motor tasks with shorter time consumed compared with VR.