Virtual Reality for Action Evaluation

TL;DR

This paper explores using low-cost VR devices combined with deep learning to evaluate physical rehabilitation exercises remotely, aiming to improve accessibility, affordability, and real-time assessment capabilities.

Contribution

It introduces a novel approach that leverages consumer-grade VR tracking data and AI for effective action evaluation in asynchronous rehabilitation.

Findings

VR tracking data can accurately evaluate rehabilitation actions

Deep learning models perform well on VR-derived data streams

Potential for accessible remote physical therapy solutions

Abstract

Physical rehabilitation plays a crucial role in restoring functional abilities, but traditional approaches often face challenges in terms of cost, accessibility, and personalized monitoring. Asynchronous physical rehabilitation has gained traction as a cost-effective and convenient alternative, but it lacks real-time monitoring and assessment capabilities. This study investigates the feasibility of using low-cost Virtual Reality (VR) devices for action evaluation in rehabilitation exercises. We leverage state-of-the-art deep learning models and evaluate their performance on three data streams (head and hands) derived from existing rehabilitation datasets that approximate VR headset and hand data. Our results demonstrate that VR tracking data can be effectively utilized for action evaluation, paving the way for more accessible and affordable remote monitoring solutions in physical therapy. By leveraging artificial intelligence techniques and consumer-grade virtual reality technology, this study proposes an approach that could potentially address some of the challenges in asynchronous rehabilitation, such as the need for expensive motion capture systems or in-person sessions.