Neural Mismatch Model for VR Sickness Quantification
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current VR sickness assessment methods are cumbersome and labor-intensive, relying on subjective questionnaires and physiological measures, which are not effective in preventing severe VR sickness and lack generalization due to limited data and resolution in stimulus evaluation.
Innovation Solution
A deep learning-based VR sickness assessment method that models the neural mismatch phenomenon using a neural network to predict VR sickness by comparing expected and actual visual signals, extracting neural mismatch features, and evaluating sickness levels based on a pre-trained neural mismatch model.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If subjective questionnaires and physiological measures are used for VR sickness assessment, then assessment can be performed, but the method becomes cumbersome and labor-intensive
Solution Approach 1:
The patent replaces the mechanical/manual system of subjective questionnaires and physiological measurements with an automated deep learning-based visual signal analysis system. The neural network automatically processes video frames to extract motion features and predict VR sickness scores, eliminating the need for manual questionnaire administration and physiological data collection while maintaining assessment accuracy
Solution Approach 2:
The patent creates a computational model that copies and simulates the human neural mismatch mechanism through deep learning. The neural network is trained to replicate how human brains process visual motion signals and generate sickness responses, allowing automated prediction of VR sickness without requiring actual human subjects or complex measurement equipment
2Measurement precision
If conventional assessment methods with limited data are used, then assessment can be performed, but generalization capability is reduced
Solution Approach 1:
The patent performs preliminary training of the deep learning model using extensive motion stimulus data before deployment. The neural network is pre-trained on a diverse dataset containing various motion patterns, frame rates, and VR content types, enabling it to generalize to new, unseen VR content without requiring additional calibration or subject testing for each new scenario
Solution Approach 2:
The patent transforms the assessment approach by changing from fixed-threshold conventional methods to a continuous probabilistic output from the neural network. The model outputs predicted VR sickness scores as continuous values based on learned patterns from training data, allowing flexible adaptation to different VR content types and motion characteristics without requiring parameter re-tuning
3Measurement precision
If deep learning-based neural mismatch modeling is used, then quantitative VR sickness prediction is achieved, but computational complexity increases
Solution Approach 1:
The patent extracts only the essential visual motion signals from VR content that are most predictive of VR sickness, rather than processing all visual information. The deep learning model selectively extracts motion features such as optical flow, motion boundaries, and temporal changes, discarding redundant information and reducing computational load while maintaining prediction accuracy
Solution Approach 2:
The patent segments the VR content processing into distinct stages: frame extraction, motion feature computation, neural network inference, and score aggregation. This segmentation allows for optimized processing at each stage, using appropriate algorithms and data structures for each task, thereby reducing overall computational complexity compared to processing the entire video stream as a single unit
Data Source
AI summary
A virtual reality (VR) sickness assessment method according to an embodiment includes receiving virtual reality content, and quantitatively evaluating virtual reality sickness for the received virtual reality content using a neural network based on a pre-trained neural mismatch model. The evaluating of the virtual reality sickness may include predicting an expected visual signal for an input visual signal of the virtual reality content based on the neural mismatch model, extracting a neural mismatch feature between the predicted expected visual signal based on the neural mismatch model and an input visual signal for a corresponding frame of the virtual reality content corresponding to the expected visual signal, and evaluating a level of the virtual reality sickness based on the neural mismatch model and the extracted neural mismatch feature.