Keywords

1 Introduction

The development of display technology and smart devices has enabled people to experience virtual reality at a very low cost. Many virtual reality glasses have been introduced in the market, but the design of most of these products lacks careful consideration of ergonomics, which results in poor usability and unpleasant user experience [1,2,3]. Though the creators of the virtual reality glasses have provided many suggestions for both developers and users, there are still many complaints about the many discomforts experienced when using virtual reality glasses. Most of the discomforts were caused by simulator sickness.

Simulator sickness is one of the major obstacles that hinder the popularity of virtual reality glasses. To solve this problem, one of the options is to simulate the human perception of the real world in virtual reality, which includes visual sense, auditory sense, tactile sense etc. Since most of the virtual reality glasses only provide visual and auditory feedback, the effects of tactile feedback, especially vibrotactile feedback, have received less attention. Hence, it is important to study the influence of vibrotactile feedback on simulator sickness, the performance of the glasses, and user satisfaction.

The aim of this study was to investigate the effects of vibrotactile feedback on simulator sickness, performance, and user satisfaction when a user wears virtual reality glasses. In addition, the study will validate the sensory conflict theory and postural instability theory since there is some conflict between these two theories. Based on the result of the experiments, this study will provide some suggestions and recommendations on the design of virtual reality glasses.

2 Literature Review

The primary cause of discomfort experienced with virtual reality glasses is simulator sickness. Though viewed as a syndrome different from motion sickness, simulator sickness is treated as a related phenomenon by many researchers because the two syndromes share many symptoms such as fatigue, headache, visual strain, sweating, nausea, etc. [4]. There are mainly four theories that explain this syndrome: sensory conflict theory, postural instability theory, eye movement theory and evolutionary theory [2]. Sensory conflict theory, the most widely accepted theory, explains that motion sickness is caused by the conflict between the visual and vestibular perceptions of human beings [5]. Virtual reality glasses cannot provide the accelerations and amplitudes equivalent to those experienced during real motion. Thus, users are likely to suffer simulator sickness in this situation. However, there is another theory explaining the cause of simulator sickness. Postural instability theory suggests that human beings, like other animals, tend to maintain their postural stability. When they are not able to balance their body posture, they are very likely to get sick [6]. In a virtual environment, human beings must learn a new strategy to control their bodies. Before they acquire the ability to maintain the balance in this new environment, they may suffer from simulator sickness. This was proved by many studies [7, 8] and this theory can also explain the adaption to the simulator as well. In eye movement theory, Ebenholtz [9, 10] suggested that two specific eye movements, optokinetic nystagmus, and vestibular ocular response, might induce some errors and thus result in motion sickness and simulator sickness. In the view of evolutionary theory, the brain of a human being regards sensory conflict as being equivalent to poisoning. The motion sickness and simulator sickness are mechanisms of self-protection, which will lead to vomiting of the poisoned food [11].

According to sensory conflict theory, providing additional motion stimulus will reduce simulator sickness. One of the options is to provide vibrotactile feedback to users wearing virtual reality glasses because it can provide more motion information, especially with respect to the amplitudes of the motion. In this way, the conflict between visual and vestibular perceptions can be reduced. However, because the vibrotactile feedback is different from real motion, from the perspective of postural instability theory, human beings cannot avoid the learning process of balancing their body in virtual reality with vibrotactile feedback. So, providing additional vibrotactile feedback cannot ensure the reduction of simulator sickness. Therefore, it is very interesting to investigate the effect of vibrotactile feedback on simulator sickness.

  • Hypothesis 1: Vibrotactile feedback will reduce simulator sickness of a user who is wearing virtual reality glasses.

Some other factors may also influence simulator sickness. The field of view influences simulator sickness. The field of view (FOV) of a single eye has a range of approximately 200° in the horizontal direction and approximately 135° in the vertical direction [12]. In the monocular system of Head-Mounted Displays (HMDs), the FOV is related to the screen size, lens size, eye relief distance, exit pupil size, and focal length. A very large FOV will lead to simulator sickness [12]. However, a narrow FOV may decrease the immersion of the user. A previous study [13] suggests that the field of view should be between 85° and 120° for virtual reality glasses. Lag and latency are other factors that may induce simulator sickness. There are three sources of lag that may influence a virtual reality system. The first source is the system lag that is influenced mainly by the speed of the central processing unit and the graphics processing unit. The second source of lag is the latency between the input device and the processor. The third source of lag is the latency between the processor and the output device. High delays may hamper the performance of the user and induce simulator sickness.

Vibrotactile feedback is mainly used in smartphones and gaming consoles. It provides users with vibrotactile information at a very low cost. The current virtual reality glasses provide visual feedback and audio feedback. Vibrotactile feedback has rarely been used in products till now. If providing vibrotactile can reduce simulator sickness, the human performance and user satisfaction may improve because users would feel more comfortable wearing virtual reality glasses. In addition, many studies suggest the positive effect of vibrotactile feedback on task performance and satisfaction. Vibrotactile cues are employed for alert, direction, spatial orientation and communication because they decrease the users’ reaction time and improve their situation awareness [14]. Studies have also shown that with vibrotactile feedback, users can enjoy a higher level of realism and immersion [15]. Based on the literature above, we can speculate that vibrotactile feedback would improve users’ performance and satisfaction. In addition, many studies have explored the usage of vibrotactile feedback in virtual reality [16, 17]. However, very few studies look into the effect of vibrotactile feedback on user experience in virtual reality, especially on simulator sickness, performance, and user satisfaction.

  • Hypothesis 2: Vibrotactile feedback will improve the users’ performance when they wear virtual reality glasses.

  • Hypothesis 3: Vibrotactile feedback will improve the users’ satisfaction when they wear virtual reality glasses.

3 Methodology

3.1 Experiment Design

The experiments were conducted on a group of participants, who received two treatments. In one treatment, the participants were asked to finish the tasks without vibrotactile feedback and in the other treatment, participants finished the tasks with vibrotactile feedback. Both treatments had visual feedback and audio feedback. To negate the learning effect, the order of the two treatments was randomized. In the experiments, participants were asked to wear virtual reality glasses and vibration vests. They used a controller that could provide vibrotactile feedback as well to play a car race game. In the game, they were asked to drive along the road as fast as they can and avoid a collision. During the experiments, the performance was recorded. After each experiment, the participants were asked to fill a simulator sickness questionnaire to assess their severity of simulator sickness. They were also asked to fill a satisfaction questionnaire to know their feelings about the virtual reality experience and vibration. After all the experiments, they were asked to fill a questionnaire to compare their relative experience in the two scenarios of vibration and no vibration.

To reduce the effect of the participants’ familiarity with the controller in the experiment, it was ensured that the task was simple. A car race game, Project CARS, was chosen. This game was of high graphic quality and compatible with virtual reality. Figure 1 shows a driving scene in the games. In the experiment, the participants were required to drive for 10 min in the game. The track in the game is 12 miles long with 95 turns.

Fig. 1.
figure 1

Driving scene in the games

Full size image

3.2 Apparatus

The VR system in the experiment requires five components: virtual reality glasses, a high-performance PC, an earphone, two controllers and a vibration vest. Figure 2 shows the set-up of the experiment including the environment and equipment. The virtual reality glasses used were Oculus Rift DK2 with a resolution of 960 × 1080 per eye.

Fig. 2.
figure 2

Set-up of experiment environment and equipment

Full size image

The configuration of the PC used for the experiment was I7 4970k and NVIDIA GTX 970, which ensured high fraps and clear image in the experiments. The controllers were two Xbox One controllers. The wires of vibration motors in one controller were cut off to remove the vibrotactile feedback because there was no such option in the game. The controller would vibrate on three occasions: speeding up, running on different grounds, and crashing with other objects. The vibration vest in the experiments was KOR-FX that vibrated according to the sound of the game. The vest would generate strong vibrations in case of loud sounds in the game, such as collisions. In one treatment, the vibrations of the vest and controller were turned on and in another treatment, the vibrations were turned off.

3.3 Measurement

There are mainly three approaches to measure simulator sickness: motion sickness checklist questionnaires, rapid self-report questionnaires, and psychophysiological measurements. Motion sickness checklist questionnaires are the most popular method to measure motion sickness. Participants are asked to assess their overall state and fill the checklist before and after the experiments. But the questionnaires usually have a long checklist that makes it difficult to fill during the experiment. Thus, rapid self-report questionnaires are developed to record the state of the participants in the process of experiments. Rapid self-report questionnaires usually have much fewer questions, but each question has more levels than checklist questionnaires. The limitation is that the data is not normally distributed, which complicates statistical analysis. Psychophysiological measurements seem preferable and objective to measure the states of participants because these methods can record the real-time information without requiring participants to pause during the experiment to report. But Lawson (2014) pointed out that psychophysiological measurements are not ideal for measurement of severity of motion sickness because: (1) the correlations between physiological data and motion sickness self-reports are usually weak and inconsistent, (2) many factors other than motion sickness can affect the nervous system, (3) physiological effects of movement or exercise can also influence the data collected in some experiments involving head movement. Hence, psychophysiological measurements are not common in studies conducted to measure the simulator sickness.

In this experiment, we use a Simulator Sickness Questionnaire (SSQ) to measure the severity of simulator sickness of the participants. SSQ is one of most popular motion sickness checklist questionnaires. It was developed by Kennedy et al. [4] and consists of 16 questions, which are related to 16 symptoms of simulator sickness. Participants usually fill out the questionnaire before and after the experiment to compare the effects of the virtual environment on the human body. However, Kennedy recommended using the post-exposure questionnaire as the index of simulator sickness instead of the difference between the pre-exposure questionnaire and the post-exposure questionnaire. The pre-exposure questionnaire can be used to test if the subjects are able to finish the experiments.

The performance was recorded based on two parameters: distance the participants drive in 10 min and the total number of errors in 10 min. In this experiment, errors included collisions, pull-ups, and retrogrades. During the experiment, the experimenters watched the screen and recorded each error. But if there was a series of collisions caused by one collision, it would be recorded as one error.

Satisfaction was measured by the questionnaire. Because there was no questionnaire to measure the satisfaction of virtual reality glasses and vibrotactile feedback, we formulated a questionnaire that covered five aspects: physical comfort, visual and auditory satisfaction, interaction, and immersion. Each aspect was assessed using several questions and the participants were asked to choose their level of satisfaction on a score of 1 to 7.

3.4 Participants

36 participants (17 males and 19 female) were recruited through social platforms but only 30 (13 males and 17 females) of them accomplished the experiment. The rest of them became too sick to finish the task. The participants were students with a bachelor’s degree or higher educational qualifications and they did not have any eye disease. The average age of the participants was 22.7 years with SD = 1.37. Among those who had the experiment, eight participants had prior experience of using virtual reality glasses, primarily for games and videos.

3.5 Procedure

This research was carried out in the Laboratory of Human Factors. The total time of the experiment was around 60 min. The participants were told the purpose of the experiments. They were asked to sign the informed consent and fill in a chart about their personal background including name, age, gender, physical condition and the previous experience using virtual reality. Then the participants were asked to wear the virtual reality glasses and vibration vest. They were taught to use the controller. They could drive in the games for a while to get used to the virtual reality scenes and operation.

According to the experiment design, the order of the two treatments for each participant was randomly disrupted. The participants put on the virtual reality glasses to finish the first task. During the task, the experimenters record the performance of the participants. After the tasks are finished, subject take off the virtual reality glasses and fill the simulator sickness questionnaire and satisfaction questionnaire. Then, they can rest for 10–20 min, which depends on their own preference. If the participants feel adequately recovered they can put on the virtual reality glasses to finish the second task. After the experiments, the subjects are interviewed.

4 Result and Analysis

The effect of vibrotactile feedback on simulator sickness was analyzed. The three sub-scores of simulator sickness were compared. Then, we examined the influence of vibrotactile feedback in virtual reality on human performance, taking the rate of error into account. The satisfaction of the participants was presented to examine the impact of vibrotactile feedback on the satisfaction. To ensure the appropriate outcome of the SSQ and Satisfaction Questionnaire, its reliability and validity should be checked. The Cronbach’s Alpha values of all the items are higher than 0.8, which ensures the reliability and validity of the questionnaire result.

SSQ scores were calculated by three sub-scores: nausea, oculomotor, and disorientation. Nausea included general discomfort, increased salivation, sweating, nausea, difficulty concentrating, stomach awareness and burping. Oculomotor included general discomfort, fatigue, headache, eye strain, difficulty focusing, difficulty concentrating and blurred vision. Disorientation included difficulty focusing, nausea, fullness of head, blurred vision, dizziness, and vertigo.

4.1 Descriptive Statistics

First, we compared the simulator sickness scores with and without vibrotactile feedback, which also included a comparison between the three sub-scores. From Fig. 3, it can be observed that nausea, oculomotor, disorientation, and simulator sickness is all higher without vibrotactile feedback. Similar comparisons were made in terms of satisfaction, as depicted in Fig. 4. However, the difference the between two treatments is not obvious. On the other hand, considering the individual difference between simulator sickness, performance, and satisfaction, a better way to compare the results is by comparing the result to themselves. Thus, we used a paired t-test in the following analysis.

Fig. 3.
figure 3

Comparison of simulator sickness scores with and without vibrotactile feedback

Full size image
Fig. 4.
figure 4

Comparison of satisfaction scores with and without vibrotactile feedback

Full size image

4.2 Effect of Vibrotactile Feedback on Simulator Sickness

To compare the mean SSQ scores obtained from the two treatments, a paired T-test is used. As shown in Table 1, simulator sickness without vibrotactile feedback (M = 15.77, SD = 8.597) is higher than with vibrotactile feedback (M = 11.73, SD = 6.992, t(30) = 3.946, p < 0.001). Three sub-scores were also investigated. Nausea was significantly higher without vibrotactile feedback (M = 4.60, SD = 2.908) than with vibrotactile feedback (M = 3.30, SD = 2.322, t(30) = 2.956, p = 0.006). Oculomotor was also significantly higher without vibrotactile feedback (M = 6.00, SD = 3.206) than with vibrotactile feedback (M = 4.33, SD = 2.893, t(30) = 3.893, p = 0.001). Disorientation was also higher without vibrotactile feedback (M = 5.17, SD = 3.445) than with vibrotactile feedback (M = 4.10, SD = 2.833, t(30) = 2.874, p = 0.008).

Table 1. Paired t-test of simulator sickness between no vibrotactile feedback and vibrotactile feedback
Full size table

4.3 Effect of Vibrotactile Feedback on Task Performance

We measured the performance in terms of the distance the participants drove and the number of errors they committed in 10 min. The distance was given in the form of percentages because the game only provided the percentage of distance covered in the total track. As is shown in Table 2, there is not a significant difference in the performance between no vibrotactile and vibrotactile feedback scenarios.

Table 2. Paired t-test of performance between no vibrotactile feedback and vibrotactile feedback
Full size table

4.4 Effect of Vibrotactile Feedback on Satisfaction

The satisfaction is measured in five aspects: wearing satisfaction, visual satisfaction, auditory satisfaction, interaction satisfaction and immersion. To compare the difference between the mean of satisfaction scores of the two treatments, a paired T-test is used. As shown in Table 3, the total score was significantly higher with vibrotactile feedback (M = 124.43, SD = 20.025) than without vibrotactile feedback (M = 120.57, SD = 20.668, t(30) = 3.147, p = 0.004). It is also suggested that interaction satisfaction with vibrotactile feedback (M = 22.37, SD = 3.917) is higher than without vibrotactile feedback (M = 21.57, SD = 4.099, t(30) = 2.658, p = 0.013). The immersion with vibrotactile feedback (M = 22.60, SD = 4.031) is significant higher than without vibrotactile feedback (M = 21.40, SD = 4.031, t(30) = 3.095, p = 0.004). No significant difference was found in wearing satisfaction, visual satisfaction, and auditory satisfaction when using virtual reality glasses without vibrotactile feedback and with vibrotactile feedback.

Table 3. Paired t-test of satisfaction between no vibrotactile feedback and vibrotactile feedback
Full size table

5 Discussion

Vibrotactile feedback can significantly reduce simulator sickness. It can be explained by Sensory conflict theory. From the literature review, we can see that the simulator sickness is caused by a mismatch between visual information and body movement. The body movement is sensed by a vestibule in the inner ear, which is the balance tense’s acceptor. Its nerve adjust function decides the balance ability. When users use virtual reality glasses, they are usually in a still position. The vestibule will inform the central nervous system that the body is still. However, the visual information suggests that the body is moving. This contradictory information will induce many symptoms. The vibrotactile feedback will reduce these symptoms by providing the movement stimulus. It may deceive the vestibule into believing that the body is in motion and thus reduce the disparity between visual information and movement information. In this way, the simulator sickness is reduced. In our experiment, the participants drove very fast in a car as their sense of their surrounding changed very fast. So, the visual information told them that they are moving fast. However, in reality, they just sit still in a chair. Without vibrotactile feedback, the participants would feel very sick. So, if we give the participants some vibrotactile feedback that would give them the perception of being in a real car, this sickness will be significantly reduced.

On the other hand, from the view of postural instability theory, this stimulus of vibration may increase of postural instability, which may, in turn, increase the simulator sickness. But the experiment result does not support this idea and it can be explained by two aspects. First, the participants were actually in a stable position and do not need to maintain the stability of posture. Second, the vibrotactile feedback only gave haptic feedback which did not affect the participant’s sense of motion.

Vibrotactile feedback can significantly improve users’ satisfaction, especially audio satisfaction, and interaction satisfaction. The vibrotactile feedback can provide users with information that cannot be offered by visual feedback and audio feedback. In other words, vibrotactile feedback will make virtual reality more real. With vibrotactile feedback, the user can feel the virtual world in a more detailed way. Due to the more information obtained by the feedback, the interaction appears more precise and interesting. In our experiment, different roads will lead to different feedbacks. For example, a grass road will generate a stronger feedback than a normal road because the grass road is rougher. But it is very difficult to see roughness on a visual system; it can only be felt from vibrotactile feedback. In the experiments, the participants are frequently made to run against a wall or other obstacles. With vibrotactile feedback, the users can feel real force when they hit something rather than just seeing it.

The performance is not significantly influenced by vibrotactile feedback. It may be the result of learning effect. The interval between the two experiments is too short and the participants usually performed much better in the second experiment. It may be much better if the participants get enough rest. In addition, some participants were not familiar with the controller operation, which also could have influenced their performance.

It was inferred from the interview after the experiment that almost all the participants preferred the vibrotactile feedback. Only one participant thought otherwise. He thought that vibrotactile feedback did not have much positive effect on him. Given that he was the participant who had the highest experience of virtual reality among our participants, this may be a very interesting finding, suggesting that the preference of virtual reality may have something to do with the extent of experience.

Another interesting finding is about the six participants who were unable to finish the task. Five of them had finished the first task but were unable to finish the second task after a rest of 20 min. According to their response, they felt more sickness, nausea, and headache in the second task than in the first task, which may suggest that the time interval between two exposures to the virtual environment may have some effect on simulator sickness and experience. It is also important to know how long it takes to recover after using virtual reality glasses.

6 Conclusion

This study examined the effect of vibrotactile feedback on simulator sickness, performance, and users’ satisfaction by conducting multiple experiments within a group of participants. The participants were asked to play a car race game in a virtual reality environment with and without vibrotactile feedback. By comparing the result of the two experiments, we arrived at two major findings: vibrotactile feedback can reduce simulator sickness in virtual reality and vibrotactile feedback can improve users’ satisfaction of virtual reality glasses. So, it is recommended to provide vibrotactile feedback during the usage of virtual reality glasses.

The major limitation of this study lies in the experiment design. The learning effect is obvious even though the participants have enough rest to reduce the symptoms of simulator sickness. Many participants performed much better in the second experiment. It should be also noted that participants in the experiments were Tsinghua University students. Their preference and acceptance of virtual reality may differ from ordinary people. Despite these limitations, we believe that this study can contribute to the design of virtual reality glasses. The simulator sickness is one of the obstacles that prevents ordinary people from accepting this new technology. Providing vibrotactile feedback may have a positive effect on simulator sickness and satisfaction. Adding a new dimension of feedback will benefit a lot of users and open a huge market for virtual reality glasses.