VR Sickness Evaluation Using Deep Learning Motion Mismatch Analysis
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for evaluating VR content-induced sickness are largely subjective and time-consuming, lacking practicality in quantitatively analyzing motion mismatch between visual and posture recognition information, which is a significant factor in causing VR sickness.
Innovation Solution
A VR content sickness evaluating apparatus using deep learning to analyze visual and posture recognition information through convolutional neural networks and long short-term memory models, generating motion mismatch maps to quantify and predict the degree of sickness.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If subjective evaluation methods (questionnaires, bio-signal measurement) are used to assess VR sickness, then evaluation comprehensiveness is improved, but time consumption and manpower requirements increase significantly
Solution Approach 1:
The patent replaces manual subjective evaluation methods with an automated deep learning-based motion mismatch analysis system. The system automatically extracts visual recognition information and posture recognition information, generates motion mismatch maps, and quantifies VR sickness without requiring human questionnaire completion or bio-signal measurement, thereby eliminating time consumption and manpower requirements while maintaining evaluation comprehensiveness
Solution Approach 2:
The system enables self-service evaluation by automatically analyzing motion mismatch between visual and posture recognition information without human intervention. The deep learning model autonomously processes VR content motion data and user posture data to generate sickness evaluation results, making the evaluation process independent of human time and effort
2Extent of automation
If deep learning-based motion mismatch analysis is implemented, then automation level is improved, but device complexity increases due to multiple analyzing units and discriminators
Solution Approach 1:
The patent segments the automation system into distinct functional modules: visual recognition analyzing unit, posture recognition analyzing unit, and discriminator. Each unit performs a specific function in the motion mismatch analysis pipeline, allowing the complex automated evaluation to be broken down into manageable components that can be implemented and maintained separately
3Measurement precision
If quantitative motion mismatch analysis is performed using deep learning, then measurement precision of VR sickness evaluation is improved, but computational requirements and processing complexity increase
Solution Approach 1:
The system performs preliminary action by pre-processing and extracting motion features from VR content and user posture data before the actual sickness evaluation. The visual recognition analyzing unit and posture recognition analyzing unit extract relevant motion information in advance, generating motion mismatch maps that are then used by the discriminator for final quantitative evaluation, thereby reducing computational burden during the critical evaluation phase
Data Source
AI summary
A VR content sickness evaluating apparatus using a deep learning analysis of a motion mismatch and a method thereof are provided. The VR content sickness evaluating apparatus analyzes a motion mismatch phenomenon between visual recognition information and posture recognition information, which occurs when a user views VR content, using deep learning and predicts and evaluates a degree of VR sickness from a difference between motion features.