Abstract
Recently, the price of HMD for virtual reality has been popularized, and the technology related to virtual reality and the market have been actively growing. But, the popularization of HMD is rapidly increasing, scientific studies on the side effects of HMD and countermeasures for improving the safety are not enough. In this paper, we propose an evaluation model that quantitatively predicts the VR sickness induced by HMD using machine learning technique. This approach will provide an evaluation model for predicting VR sickness caused by contents for HMD objectively and in real time without any additional user experiment. We expect this approach to support the stability and quality improvement of VR content based on HMD.
You have full access to this open access chapter, Download conference paper PDF
1 Introduction
Recent advances in hardware technology have led to the production of consumer appropriate head-mounted displays (HMDs) such as the Oculus Rift, which is theoretically ideal for personal use in immersive virtual reality (VR) applications including gaming, simulation, and film. As the virtual reality is expected to converge with other industries, worldwide sales of HMDs will increase to 39.8 million units by 2018 and the number of users will reach 171 million. Although the popularization of HMD is rapidly increasing, scientific studies on the side effects of HMD and countermeasures for improving the safety are not enough.
Most VE-related sickness definitions describe it as a type of motion sickness. According to Burdea and Coiffet, cyber sickness is a form of motion sickness that results from interaction with or immersion in VEs [1]. visually-induced motion sickness (VIMS) refers to sickness experiences evoked by immersion in computer-generated virtual environments without the use of mechanical simulators. [2,3,4] Because of the lack of real motion in VIMS, it has been suggested that stimuli-related features such as vection, lag, and image quality are the main contributors to VIMS symptoms [5].
In addition to the direct symptoms just listed, several other phenomena are closely associated with motion and VR sickness, and potentially persist long after usage. One of the most troubling aspects of VR sickness is that symptoms might last for hours or even days after usage [6] (Fig. 1).
2 Problem
HMD provides 360-degree stereoscopic image to the user using head tracking and wide viewing angle display technology, while blocking real-world perception and perception. This maximizes the immersion and realism of the user. But, the problem of VR sickness is expected to be serious. VR sickness is one of the physiological side effects that occur when experiencing a virtual environment, and it shows symptoms like real-life motion sickness such as dizziness, eye fatigue, and nausea. Negative experiences such as VR sickness may lead to aversion learning to the virtual environment and may result in users distancing access to the virtual environment. Therefore, the study on VR sickness is essential for the expansion and quality improvement of virtual reality related business using HMD in the future (Fig. 2).
In this paper, we propose an evaluation model that quantitatively predicts the VR sickness induced by HMD using machine learning technique. This approach will provide an evaluation model for predicting VR sickness caused by contents for HMD objectively and in real time without any additional user experiment.
3 Main Idea
For this purpose, we will study a model for predicting VR sickness using semi-supervised learning techniques, which is a class of machine learning using both labeled and unlabeled data for training. Typically, a small amount of labeled data with a large amount of unlabeled data. We perceive the relative motion between the camera and the subject as the main characteristic of the VR sickness due to the nature of the HMD that provides the first-person view. In order to extract these features from the image, we introduce optical flow, which is the pattern of apparent motion of objects, surfaces, and edges in a visual scene caused by the relative motion between an observer and a scene.
And, the desired output value for a given input data is classified into five levels according to the degree of comfort. For labeled training data, experiments are performed to measure VR sickness for subjects. That is, the subjects are stimulated by using the same image as the machine learning, and the subjects’ reaction are analyzed through subjective self-report and physiological measurement. The basic concept of physiological measurement of VR sickness is that a psychological change causes a physiological change, and a change pattern of a physiological response changes according to a specific psychological state (Fig. 3).
4 Implementation
We used NoLimit2 to generate roller coaster content to create stimuli for HMD-based virtual environments. The user wears an HMD and rides on a virtual roller coaster track. At this time, we store the image at the point of view of the user. And we use openCV’s optical flow algorithm to extract features from this image. Machine learning is performed using the information extracted from the image and the motion estimation information acquired through the user experiment. At present, the core technology for machine learning has been secured in relation to this research, and systematic experiments will be conducted based on this.
5 Conclusion
The prediction model proposed in this paper will not only quantify the VR sickness caused by the HMD contents without additional user experiments, but also predict in real time. The predicted VR sickness is an objective criterion for users to choose the content suitable for them and will be used as a guideline for developing VR contents to developers. We expect this approach to support the stability and quality improvement of VR content based on HMD.
References
Burdea, G., Coiffet, P.: Virtual Reality Technology. Wiley, Hoboken (2003)
Regan, E.C., Price, K.R.: The frequency of occurrence and severity of side-effects of immersion virtual reality. Aviat. Space Environ. Med. 65, 527–530 (1994)
Ellis, S.R.: Nature and origins of virtual environments: a bibliographic essay. Comput. Syst. Eng. 2, 321–347 (1991)
Howarth, P.A., Hodder, S.G.: Characteristics of habituation to motion in a virtual environment. Displays 29, 117–123 (2008)
Stanney, K.M., Mourant, R.R., Kennedy, R.S.: Human factors issues in virtual environments: a review of the literature. Presence: Teleoperators Virtual Environ. 7, 327–351 (1998)
Stanney, K.M., Kennedy, R.S.: Aftereffects from virtual environment exposure: how long do they last? In: Proceedings of the Human Factors and Ergonomics Society Annual Meeting, vol. 48, no. 2, pp. 1476–1480 (1998)
Acknowledgments
This work was supported by the National Research Foundation of Korea Grant funded by the Korean Government (NRF-2012S1A5A2 A03034747).
Author information
Authors and Affiliations
Corresponding author
Editor information
Editors and Affiliations
Rights and permissions
Copyright information
© 2017 Springer International Publishing AG
About this paper
Cite this paper
Kim, J.B., Park, C. (2017). A Study on VR Sickness Prediction of HMD Contents Using Machine Learning Technique. In: Stephanidis, C. (eds) HCI International 2017 – Posters' Extended Abstracts. HCI 2017. Communications in Computer and Information Science, vol 714. Springer, Cham. https://doi.org/10.1007/978-3-319-58753-0_6
Download citation
DOI: https://doi.org/10.1007/978-3-319-58753-0_6
Published:
Publisher Name: Springer, Cham
Print ISBN: 978-3-319-58752-3
Online ISBN: 978-3-319-58753-0
eBook Packages: Computer ScienceComputer Science (R0)