Movement tracking accuracy of HTC Vive VR System during upper body discrete motion tasks

Virtual reality (VR) technology has expanded its potential use beyond gaming and into the medical space. While results from studies using VR to improve patient outcomes in rehabilitation are encouraging, there is a paucity of studies examining the accuracy of the motion capture data provided by the VR systems. This leads to gaps in the validation evidence for such VR systems that purport to track human movement for medical applications. Therefore, the goal of this study was to conduct an experiment that examines the accuracy of a VR-based motion capture (MoCAP) system compared to a traditional optoelectronic motion capture system as human participants perform a variety of motions based on therapeutic outcome measures. Fourteen subjects (4 females, average age of all subjects 37 ± 11 years) completed a series of discrete motion tasks typically performed in physical therapy. These subjects wore hardware from the HTC Vive Pro that was marked with optical markers tracked by OptiTrack Flex 13 cameras. A custom script collected data from the Vive and Motive software collected the data for the Flex 13 cameras. Bland-Altman (B-A) plots were generated for the range of motion of the motion tasks. Root mean square error (RMSE) was calculated between the two systems across the range of motion. The B-A analysis showed that there was strong agreement of the range of motion of the subjects across all discrete motion tasks. RMSE values varied across the motion tasks where values were generally below 37 mm, 43 mm, and 29 mm for the X, Y, Z axes respectively. These results show that, in the case of the HTC Vive, it is accurate at tracking human motions that would be performed in physical therapy. This information along with the study described below can be applied to other VR systems to standardize how VR MoCAP accuracy is assessed.