Mobile Virtual Reality Systems Research Articles

Enabling Mobile Virtual Reality with Open 5G, Fog Computing and Reinforcement Learning

Virtual reality (VR) is an emerging technology reshaping interactive experience and can be widely applied in gaming, operation training, and so on. Despite its great potentials, most existing VR solutions suffer from low mobility support, high latency, quick battery drain, as well as high cost of user devices. To overcome these issues, a mobile VR system is designed and implemented. The basic idea is to take advantage of 5G and fog computing to realize high bandwidth and low latency VR service. Meanwhile, our design features the integration of an open 5G base station (BS) and an intelligent controller. With the help of artificial intelligence (AI) and the interfaces provided by the BS and fog VR servers, the controller can wisely adjust both the BS-level and application-level parameters to enhance system performance. Using our prototype system, the impact of various parameters, the superiority of fog computing over cloud computing in supporting VR, and the effectiveness of AI in optimizing system performance are all demonstrated. In addition, multi-dimensional resource optimization for VR delivery and strategy design for VR service migration are identified as two promising future research directions.

Evaluation of Visual Acuity Measurement Based on the Mobile Virtual Reality Device

Job-related vision standards have become an increasing concern in recent years. Mobile visual acuity measurements enable early detection and diagnosis of visual impairments and are being used around the world. However, the reliability of mobile visual acuity testing has not yet been fully demonstrated. A simple virtual reality (VR) system combining a mobile phone and a VR cardboard device has the potential as a reliable visual acuity evaluation system due to its fully controlled environment. Visual acuity measurements taken via this type of VR system were evaluated by comparing them with those obtained using the traditional Snellen chart. This study gathered data according to different parameters, including right or left eye, with or without corrective vision devices, and the learning effects of the system. The results showed that the VR system had an accuracy of up to 96.43% and 92.86% for the left and right eyes, respectively, for participants not using corrective devices. In the same group, the proposed system provided significant correlation results for Spearman’s r parameters for the left and right eyes (0.7342 and 0.8188, respectively), as compared to those obtained using a traditional approach. Therefore, despite some limitations, a mobile VR system has potential as a self-diagnostic tool for rapid, low-cost visual acuity measurements in a fully controlled environment as well as for providing historical vision data and tracing for the early detection of visual impairments or conditions.

Analyzing the Effect of Masking Length Distribution of MLM: An Evaluation Framework and Case Study on Chinese MRC Datasets

Machine reading comprehension (MRC) is a challenging natural language processing (NLP) task. It has a wide application potential in the fields of question answering robots, human-computer interactions in mobile virtual reality systems, etc. Recently, the emergence of pretrained models (PTMs) has brought this research field into a new era, in which the training objective plays a key role. The masked language model (MLM) is a self-supervised training objective widely used in various PTMs. With the development of training objectives, many variants of MLM have been proposed, such as whole word masking, entity masking, phrase masking, and span masking. In different MLMs, the length of the masked tokens is different. Similarly, in different machine reading comprehension tasks, the length of the answer is also different, and the answer is often a word, phrase, or sentence. Thus, in MRC tasks with different answer lengths, whether the length of MLM is related to performance is a question worth studying. If this hypothesis is true, it can guide us on how to pretrain the MLM with a relatively suitable mask length distribution for MRC tasks. In this paper, we try to uncover how much of MLM’s success in the machine reading comprehension tasks comes from the correlation between masking length distribution and answer length in the MRC dataset. In order to address this issue, herein, (1) we propose four MRC tasks with different answer length distributions, namely, the short span extraction task, long span extraction task, short multiple-choice cloze task, and long multiple-choice cloze task; (2) four Chinese MRC datasets are created for these tasks; (3) we also have pretrained four masked language models according to the answer length distributions of these datasets; and (4) ablation experiments are conducted on the datasets to verify our hypothesis. The experimental results demonstrate that our hypothesis is true. On four different machine reading comprehension datasets, the performance of the model with correlation length distribution surpasses the model without correlation.

Usefulness of the Mobile Virtual Reality Self-Training for Overcoming a Fear of Heights

A mobile virtual reality system, equipped with built-in variables such as heart rate (HR), gaze-down data, and subjective fear rating, can allow individuals with a fear of heights to overcome it by self-training. This study aimed to verify the efficacy and safety of the training program. Forty-eight volunteers completed the four-session self-training program over 2 weeks. They were allocated into either low- or high-fear group by the Acrophobia Questionnaire (AQ)-anxiety scores, and then the changes of the built-in variables and AQ-anxiety scores were analyzed between the groups. The safety was assessed using the Simulator Sickness Questionnaire (SSQ). The AQ-anxiety scores were significantly decreased after self-training in both groups, and the degree of decrease was significantly greater in the high-fear group than in the low-fear group. Gaze-down percentage and subjective fear rating showed a significant group difference, but HR did not. The SSQ scores were within the acceptable level. These results suggest that the training effect was greater in the high-fear group than in the low-fear group. This mobile program may be safely applicable to self-training for individuals with high scores on the fear of heights by repeated exposure to virtual environments with the embedded feedback system.

Subjective visual vertical assessment with mobile virtual reality system

Background and objectiveThe subjective visual vertical (SVV) is a measure of a subject's perceived verticality, and a sensitive test of vestibular dysfunction. Despite this, and consequent upon technical and logistical limitations, SVV has not entered mainstream clinical practice. The aim of the study was to develop a mobile virtual reality based system for SVV test, evaluate the suitability of different controllers and assess the system's usability in practical settings. Materials and methodsIn this study, we describe a novel virtual reality based system that has been developed to test SVV using integrated software and hardware, and report normative values across healthy population. Participants wore a mobile virtual reality headset in order to observe a 3D stimulus presented across separate conditions – static, dynamic and an immersive real-world (“boat in the sea”) SVV tests. The virtual reality environment was controlled by the tester using a Bluetooth connected controllers. Participants controlled the movement of a vertical arrow using either a gesture control armband or a general-purpose gamepad, to indicate perceived verticality. We wanted to compare 2 different methods for object control in the system, determine normal values and compare them with literature data, to evaluate the developed system with the help of the system usability scale questionnaire and evaluate possible virtually induced dizziness with the help of subjective visual analog scale. ResultsThere were no statistically significant differences in SVV values during static, dynamic and virtual reality stimulus conditions, obtained using the two different controllers and the results are compared to those previously reported in the literature using alternative methodologies. The SUS scores for the system were high, with a median of 82.5 for the Myo controller and of 95.0 for the Gamepad controller, representing a statistically significant difference between the two controllers (P<0.01). The median of virtual reality-induced dizziness for both devices was 0.7. ConclusionsThe mobile virtual reality based system for implementation of subjective visual vertical test, is accurate and applicable in the clinical environment. The gamepad-based virtual object control method was preferred by the users. The tests were well tolerated with low dizziness scores in the majority of patients.