Chinese

2019,  1 (3):   265 - 275   Published Date：2019-6-20

DOI: 10.3724/SP.J.2096-5796.2019.0009

Abstract

Background
Ultrasonic tactile representation utilizes focused ultrasound to create tactile sensations on the bare skin of a user’s hand that is not in contact with a device. This study is a preliminary investigation on whether different ultrasonic haptic rendering methods have an impact on the perceptual threshold.
Methods
This study conducted experiments with the adaptive step method to obtain participants’ perceptual thresholds. We examine (1) whether different parts on the palm of the hand have different perceptual thresholds; (2) whether the perceptual threshold is different when the ultrasonic focus point is stationary and when it moves in different trajectories; (3) whether different moving speeds of the ultrasonic focus point have an influence on the perceptual threshold; and (4) whether the addition of a DC offset to the modulating wave has an impact on the perceptual threshold.
Results
The results show that the center of the palm is more sensitive to ultrasonic haptics than the fingertip; compared with a fast-moving focus point, the palm is more sensitive to a stationary and slow-moving focus point. When the modulating wave has a DC offset, the palm is sensitive to a much smaller modulation amplitude.
Conclusion
For the future ultrasonic tactile representation systems, dynamic adjustment of intensity is required to compensate the difference in perceptual thresholds under different rendering methods to achieve more realistic ultrasonic haptics.

Content

1 Introduction
In the fields of virtual reality and human-computer interaction, humans mainly rely on visual feedback and auditory feedback to obtain information. Visual feedback is obtained through display, and auditory feedback is presented through headphones, speakers, or other devices. In addition to audiovisual feedback, haptic feedback, such as phone vibration and gamepad vibration, is an important type of feedback and has been widely applied in human-computer interaction. Haptic feedback is usually based on haptic rendering, which refers to the process of applying the desired tactile sensory stimuli to the user and conveying the tactile information of the virtual object, enabling the user to “touch” the virtual object. Adding tactile representation into the field of human-computer interaction will improve the quality of interaction with the virtual world and help us obtain more information.
At present, in addition to vibrating tactile representation[1], which has been widely used, emerging tactile representation technologies include air film tactile representation[2], electrostatic tactile representation[3], and ultrasonic tactile representation. Unlike the other two aforementioned technologies, which use a touch screen or a wearable device, ultrasonic tactile representation is a noncontact haptic rendering technology that creates tactile sensations in free space using an ultrasound transducer array and an ultrasonic haptic rendering algorithm. Ultrasonic haptic rendering utilizes ultrasonic focusing technology and modulation to apply the desired tactile sensory stimuli to a certain point in mid-air by controlling the phase and intensity of the ultrasound pulses emitted by each ultrasound transducer. When focused on a user’s skin surface, ultrasound produces tactile sensation that can be perceived by the user. Typical ultrasonic rendering methods include amplitude modulation and spatiotemporal modulation[4,5].
Methods for rendering ultrasonic haptics in a space of higher dimension, in a space of larger scale, and in inhomogeneous mediums have been investigated in previous studies[6－9]. In tactile human-computer interaction, to offer realistic and effective tactile sensations to users, it is important to investigate human subjective feelings in addition to rendering methods. Tactile sensations can be perceived by stimulating mechanoreceptors. Owing to the distribution density and characteristics of mechanoreceptors on the skin of a user’s hand, tactile sensations in different parts of the hand are different for the same tactile stimulus[10]. The following questions thus arise: Is this also true for ultrasonic tactile stimulus? Will tactile stimuli produced by ultrasound focus points of different moving speeds or different moving trajectories also have an effect on tactile sensations? If such effects exist, when rendering the tactile sensation of a particular haptic effect, it is necessary to consider these influential factors, and correspondingly adjust the intensity of ultrasound in the tactile rendering methods to generate more realistic tactile sensations and improve the quality of the tactile interactive experiences.
This study is a preliminary investigation into the relationship between different ultrasound haptic rendering methods and user’s perceptual thresholds. The study mainly includes the following aspects: whether the perceptual thresholds of different parts of the hand to the ultrasonic haptics is different, whether different moving trajectories of a focus point and the different moving speeds of one or more focus points have an influence on the perceptual thresholds of the human hand, and whether the addition of a DC offset to the modulating wave has an influence on the perceptual threshold. The results could be valuable for the design of future ultrasonic tactile representation systems because dynamic adjustment of intensity is required to compensate the difference in perceptual thresholds under different rendering methods to achieve more realistic ultrasonic haptics.
The remainder of this paper is structured as follows: The second section introduces the related works in the field of ultrasonic haptics. The third section presents the principle of ultrasonic haptics. In the fourth section, experiments are conducted to explore perceptual thresholds of the hand under different conditions. The experimental data are presented and analyzed in the fifth and sixth sections, respectively. The final section summarizes the conclusions of this study.
2 Related works
Many studies related to the ultrasonic tactile representation technology have been conducted. Iwamoto et al. used a hexagonal annular array consisting of 91 ultrasound transducers to produce a fixed ultrasound focus in the one-dimensional space of the hexagonal central axis[6]. Hoshi et al. improved this ultrasound array by constructing a square array of 324 ultrasound transducers that produced movable focal points in 3D free space, and the phase and intensity of each transducer could be individually controlled[7]. The ultrasound can not only propagate and be focused through the homogeneous medium, but also be focused through the inhomogeneous medium. Inoue et al. used a three-dimensional surrounded ultrasound array to render and reconstruct the pressure field in an inhomogeneous field by using the time reversal and finite element methods[8]. Spelmezan et al. studied tactile sensation generated by the propagation of ultrasound through the user’s palm[9]. The palm is an inhomogeneous medium, and ultrasound is distorted during its propagation. By using time-reversed signal processing techniques, the distorted ultrasound is focused on a fixed point on the skin, which produces tactile sensations. Hasegawa and Shinoda[11] presented a multiunit ultrasound phased array that produced tactile sensations 1m away from the surface of the array, greatly increasing the distance of haptic rendering. Kervegant et al. demonstrated a human-computer interaction experience that combined smart glasses and a haptic device[12]. A hybrid holographic display and ultrasound transducer array provide realistic and tangible feedback to a hologram. This feedback enhances the presence of objects and provides a reference for new methods of interaction between people and virtual environments.
Long et al. proposed a method for rendering volumetric haptic shapes in mid-air using focused ultrasound and evaluated the effectiveness of this method to convey shape information by using a user’s subjective evaluation[13]. Frier et al. used spatiotemporal modulation to study the influence of the moving speed of the focus point on the user’s tactile perception, and concluded that the hand’s perceptual intensity peaked when the moving speed of the focus point was 5—8ms-1[5]. In addition to spatiotemporal modulation, it is possible to generate tactile sensations using amplitude modulation. Obrist et al. studied the description of the tactile experience by the participants with the modulation frequencies of 16Hz and 250Hz[4]. According to the classification of the study results, 14 categories for a human-experiential vocabulary were proposed and they were connected with the physiological and neurophysiological data of human hands. Hasegawa and Shinoda[11] detected the sensitive threshold for frequency from the perspective of user’s subjective perception and concluded that the minimum sensitive threshold is 200Hz using the sinusoidal modulation wave.
3 Ultrasonic haptic based on amplitude modulation
Ultrasonic haptic feedback is based on an array of ultrasound transducers, as shown in Figure 1. The ultrasound pulses emitted by each ultrasound transducer reach the target point at the same time, and they have the same phase at the target point, thereby generating pressure and pushing the propagation medium to move in one direction. This is called ultrasound focusing. If skin is located at the target point, the mechanoreceptors of the skin can be stimulated to produce tactile sensations, as illustrated in Figure 2.
In detail, the propagation time of ultrasound pulses from each transducer to the focus point is given by the distance from the transducer to the focus point divided by the propagation speed of ultrasound pulses. The delay of each ultrasound transducer emitting a pulse of ultrasound is controlled according to the propagation time of ultrasound pulses from the transducer to the focus point, so that ultrasound transducers farther from the target point emit ultrasound pulses earlier, and those closer to the target point emit ultrasound pulses later; thus, all the ultrasound pulses emitted by transducers simultaneously reach the target point and are focused.
When ultrasound pulses with constant intensity and frequency (Figure 3a) are focused, the pressure at the focus point is constant, and the skin does not perceive tactile sensations. Hence, it is necessary to modulate high-frequency ultrasound with a lower-frequency signal. Modulation changes the intensity of the ultrasound periodically, and the pressure at the focus point also changes periodically; hence, the skin can perceive tactile sensations. In this study, amplitude modulation is used, and a sine wave with the frequency of 200Hz is used as the modulating wave. Assuming that the DC offset of the modulating wave is α, the peak-to-peak amplitude of the modulating wave is β, and the modulating wave is
$F t = β 2 ( 1 - c o s 2 π t ) + α$
When α=0, the minimum value of the modulating wave is 0. When α>0, the minimum value of the modulating wave is not 0, which is equivalent to adding a DC offset greater than zero to the modulating wave. In this study, α and β are normalized, i.e., the maximum ultrasound intensity that can be outputted by the experiment hardware is defined as 1.0. As examples of the modulated wave, when β=0.3 and α=0, the modulated waveform is shown in Figure 3b; when β=0.3 and α=0.7, the modulated waveform is shown in Figure 3c.
4 Experiments
We conducted experiments to examine whether different parts of the palm of the hand have different perceptual thresholds, whether the user’s perceptual threshold is different when the ultrasonic focus point is stationary and when it moves in different trajectories, whether different moving speeds of one or more ultrasonic focus points have an influence on the user’s perceptual threshold, and whether the addition of a DC offset to the modulating wave influences the perception threshold.
4.1　Apparatus
The experiment used Ultrahaptics’ USX ultrasound transducer array, which has 256 ultrasound transducers arranged in a square space of 16cm × 16cm. Each ultrasound transducer emits ultrasound pulses of 40 kHz and has a diameter of 10.3mm. A piece of foamed plastic was placed above the array at a fixed relative position, as shown in Figure 1. During the experiment, the participants placed their hand on the foamed plastic with the palm facing down, thereby ensuring that the hand was at a fixed position relative to the transducer array and the bottom plane of the hand was close to parallel to the upper surface of the ultrasound array. The optimal area for rendering ultrasound haptics using the transducer array is between 10cm and 30cm above the surface of the array. When participants placed their hand on the foamed plastic as required by the experiment, the lower surface of the hand was measured to be approximately 19cm from the upper surface of the array; hence, in each of the experiments, the location of the ultrasonic focus points was set on a plane 19 cm above the upper surface of the array, which was within the optimal area.
The signal for driving the ultrasound transducers is a rectangular wave of 40kHz with variable duty ratio. The intensity of the ultrasound is controlled by the duty ratio of the voltage waveform supplied to each ultrasound transducer. When the duty ratio is 0, i.e., the voltage is a constant of 0V, no ultrasound is emitted, and the intensity of ultrasound at the focus point is 0. At this time, the peak-to-peak amplitude of the corresponding modulating wave is β=0, and the DC offset α=0. When the signal with the maximum duty ratio (3/25) is continuously delivered, the intensity of the ultrasound becomes the strongest, but there are no tactile sensations. At this time, the peak-to-peak amplitude of the corresponding modulating wave is β=0, and the DC offset α=1. When the duty ratio of the signals varies periodically between zero and the maximum value, the intensity of the ultrasound at the focus point also varies periodically, the tactile sensation is the strongest, and the peak-to-peak amplitude of the corresponding modulating wave is β=1 with the DC offset α=0.
4.2　Experimental procedure
Each participant was involved in 11 experiments and each experiment obtained a tactile perceptual threshold under a certain condition. The participants were required to focus their attention during the experiment; hence, to avoid mental fatigue of participants, the experiments were not conducted continuously. The 11 experiments were divided into small experiment blocks, each containing 2 to 3 experiments. After each participant completed an experimental block, he/she took a break and then proceeded to the next experiment block. The order and conditions of the 11 experiments are shown in Table 1.
Experimental index and experimental conditions
Experiment Block Experiment Index Experiment Conditions
1 1 α > 0, the focus point moves in a circle, 1 cycle per second
1 2 α = 0, the focus point moves in a circle, 1 cycle per second
2 3 The focus point moves in a line
2 4 The focus point is fixed at the center of the palm
3 5 The focus point is fixed at the tip of the middle finger
3 6 The focus point is fixed at the root of the middle finger
4 7 The focus point moves in a circle, 3 cycles per second
4 8 The focus point moves in a circle, 1 cycle per 3 seconds
4 9 The focus point moves in a circle, 3 cycles per second
5 10 Four focus points move in a circle, 1 cycle per 3 seconds
5 11 Four focus points move in a circle, 1 cycle per second
First, the perceptual threshold of different parts of the hand was examined, and the corresponding experiment index was 4, 5, and 6, as shown in Table 1. Second, it was investigated whether the perceptual threshold of the palm was different for moving trajectories of the focus point, and the corresponding experiment index was 2, 3, and 4, as shown in Table 1. Third, it was studied whether different moving speeds of a focus point influenced the perceptual threshold of the palm, and the corresponding experiment index was 2, 7, and 8, as shown in Table 1. Fourth, it was investigated whether the moving speeds of multiple focus points influenced the perceptual threshold of the palm. We generated four focus points simultaneously in the experiments. The four focus points simultaneously moved in the same direction and at the same moving speed, and the corresponding experiment index was 10 and 11, as shown in Table 1. Finally, the perceptual threshold of the palm was detected with and without the addition of a DC offset to the modulating wave. The corresponding experiment index was 1 and 2, as shown in Table 1. In the first experiment, α>0, β was varied between 0 and 1, and the maximum value of the modulating wave was fixed at 1; in the second experiment, α=0 and β was varied between 0 and 1. In the second, third, fourth, and fifth experiment blocks, the sine wave with α=0 and β varying between 0 and 1 was used as the modulating wave. During all the experiments, the line was formed by a reciprocating movement of the focus point whose speed changes sinusoidally. The length of the line was 3 cm, and the radius of the circle was 3 cm.
Twelve participants (6 males and 6 females, ages 22—31 years) took part in the experiment. Each experiment used the adaptive step method[14] to detect the perceptual threshold. In the adaptive step method, participants attempt to differentiate a group of tactile sensations presented to them. After differentiating a group of tactile sensations, a new group with a different differentiating difficulty is presented to them. If the participants differentiate the group correctly, the difficulty of the next group is increased, whereas if the participants differentiate the group wrongly, the difficulty of the next group is decreased. As the experiment progresses, the difficulty of differentiation will gradually converge to a certain level, which represents the perceptual threshold.
In detail, each group consisted of three haptic effects, labeled A, B, and C. Two of the haptic effects were blank, i.e., the peak-to-peak amplitude β of the modulating wave was 0 and the participant could not feel the ultrasonic tactile sensations. The other haptic effect was nonblank and had a β between 0 and 1; the participant could feel the ultrasonic tactile sensations when β was sufficiently large. The three haptic effects were labeled in random order.
The participants switched among the three haptic effects A, B, and C by pressing the space bar of a keyboard, and the letter A, B, or C corresponding to the current haptic effect was displayed on the computer screen, so that the participants could know the label of the haptic effect currently being presented. The participants switched among the effects repeatedly at will and there was no time limit. The participants selected the effect that they thought was nonblank according to their subjective feeling, by pressing the keys A, B, or C on the keyboard. When the participants chose correctly, the β value corresponding to the nonblank effect in the next group would decrease by λ. When the participants chose wrongly, the β value corresponding to the nonblank effect in the next group would increase by λ. At the starting point of each experiment, a large β value that could produce a strong tactile sensation was used, which could help the participants adapt to the experiment process, and the initial value of λ was 0.2. When β decreased after a group and then increased after the next group, or vice versa, it was called a reversal. After every three reversals, the value of λ was halved. When λ was equal to 0.125, the experiment ended. A larger λ value could cause β to approach the threshold more quickly, and a smaller λ value could be used to estimate the threshold more accurately. Each experiment consisted of 15 to 35 groups. The average β value of the last five groups was the estimated threshold.
During the experiment, the ultrasound transducer array emitted an audible sound. When β was different, the volume of the sound was also different. Therefore, the sound would provide a hint to the participants regarding the current haptic effect if they could hear it. We asked the participants to wear headphones while listening to the pre-recorded transducer array noise combined with natural rain sound, and hence, they could not hear the varying sound emitted by the transducer array.
5 Experimental results
5.1　Different parts of the hand
A fixed focus point is generated at the center of the palm, the root of the middle finger, and the tip of the middle finger. The specific parts are shown in Figure 4. The mean and standard deviation of the perceptual thresholds at different parts of the hand from 12 participants are shown in Figure 5. The p-values of a paired t-test of the perceptual thresholds of different hand parts are shown in Table 2.
Paired t-test analysis results of perceptual thresholds of different parts of the hand
Center of palm vs. root of middle finger Center of palm vs. tip of middle finger Root of middle finger vs. tip of middle finger
p-value 0.60 0.18 0.04
5.2　Moving trajectories of the focus point
The moving trajectories of the focus point are shown in Figure 6, including a fixed focus point, a focus point reciprocating in a line, and a focus point moving in a circle. For different moving trajectories, the mean and standard deviation of the perceptual thresholds from 12 participants are shown in Figure 7. The p-values of a paired t-test of different trajectories are shown in Table 3.
Paired t-test analysis results of different moving trajectories of a focus point
Fixed focus point vs. circle Fixed focus point vs. line Line vs. circle
p-value 0.14 0.28 0.31
5.3　Moving speeds of the focus point
The moving trajectory of the focus point is a circle, and the moving speeds are 3 cycles per second, 1 cycle per 3 seconds, and 1 cycle per second. For different moving speeds, the mean and standard deviation of the perceptual threshold from 12 participants are shown in Figure 8. The p-values of a paired t-test for different moving speeds of the focus point are shown in Table 4.
Paired t-test analysis results of different moving speeds of a focus point
3 cycles per second vs. 1 cycle per 3 seconds 3 cycles per second vs. 1 cycle per second 1 cycle per 3 seconds vs. 1 cycle per second
p-value 0.06 0.15 0.39
5.4　Moving speeds of multiple focal points
The four focus points are on the four vertices of a square and move in the circumscribed circle of the square, and the moving speeds are 1 cycle per 3 seconds and 1 cycle per second. For different moving speeds, the mean and standard deviation of the perceptual thresholds from 12 participants are shown in Figure 9. The p-value of a paired t-test for the two moving speeds is 0.15.
5.5　With and without adding DC offset to modulating wave
The DC offset α of the first modulating wave is 0, the value of troughs is fixed to 0, and β gradually increases from 0. For the second modulating wave, α is greater than 0, the value of crests is fixed to 1.0, and β also increases from 0 gradually. The perceptual threshold is the value of β that can be perceived by the palm. The moving trajectory of the focus point is a circle, and the moving speed is 1 cycle per second. For different modulating waves, the mean and standard deviation of the perceptual thresholds of the 12 participants are shown in Figure 10.
5.6　Fatigue factors
Owing to the large number of experiments and the long duration of each experiment, the participants were prone to fatigue. In each experiment block, each participant was required to participate in 2 to 3 experiments. We analyze whether the fatigue factors will affect the experimental results in one of the experiment blocks. In that block, each participant participated in 3 experiments, in which the first and third experiments were in the same experimental condition, i.e., the focus point moved in a circular trajectory on the palm, and the moving speed was 3 cycles per second. The mean and standard deviation of the perceptual thresholds from the 12 participants are shown in Figure 11. The p-value of a paired t-test of the two experiments is 0.47.
6 Analysis
The perceptual thresholds of the center of the palm, the root of the middle finger, and the tip of the middle finger were 0.34, 0.37, and 0.43, respectively. The tip of the middle finger was less sensitive, and the center of the palm was more sensitive. From the results of the paired t-test, the difference between the perceptual thresholds on the tip of the middle finger and the root of the middle finger was significant (p=0.04).
When the moving trajectories were a fixed focus point, a line, and a circle, the perceptual thresholds of the palm were 0.34, 0.31, and 0.29, respectively; the palm was relatively more sensitive to the focus points moving in a circle and a line, whereas its sensitivity to a fixed focus point was relatively low. The palm appeared to be more sensitive to the moving focus point, whereas this was not verified in the results of the paired t-test analysis (p=0.14).
The perceptual thresholds were 0.32, 0.27, and 0.29 when the moving speeds of the focus point were 3 cycles per second, 1 cycle per 3 seconds, and 1 cycle per second, respectively. The moving speed had a significant influence on the perceptual threshold of the hand (p=0.06).
When the moving speeds of the four focus points were 1 cycle per 3 seconds and 1 cycle per second, the perceptual thresholds of the palm were 0.53 and 0.50, respectively. No significant influence of different moving speeds of multiple focus points was observed (p=0.15). As the ultrasound intensity of a focus point and that of multiple focus points rendered by the ultrasound transducer array used in the experiments are not comparable, it is not possible to compare the perceptual thresholds of one and multiple focus points.
When there was no DC offset in the modulating wave (α=0), the perceptual threshold of the palm was 0.29, i.e., the palm could perceive a tactile sensation when β was at least equal to 0.29. When a DC offset was added to the modulating wave (α>0) so that the peak intensity of the modulating wave remained fixed at 1.0, the palm could perceive tactile sensation when β was as small as 0.026. From the experimental results, it can be observed that, if we increase the peak-to-peak amplitude of the modulating wave from zero, we require a much smaller amplitude to generate a tactile sensation on the user’s hand when starting with a DC offset to the modulating wave, compared with when starting without a DC offset to the modulating wave. Among the 12 participants, 4 participants claimed that they had felt a slight breeze on their hand when α=1 and β=0, whereas the other 8 participants claimed to experience no such feeling.
The experiment conducted twice under identical conditions resulted in perceptual thresholds of 0.32 and 0.34. The influence of fatigue factors on the experimental results is not significant (p=0.47).
During the experiments, when β approached the perceptual threshold, the subjects could not recognize the shape of the moving trajectory of the focus point. Among the three moving trajectories of the focus point, it was claimed by the subjects that the circular trajectory was easier to recognize than the other two.
7 Conclusion
In this study, the sensitivities of the human hand to ultrasonic haptics under different rendering methods have been investigated, including the perceptual thresholds of different parts of the hand, the perceptual thresholds of the palm for different moving trajectories and moving speeds of focus points, and the perceptual thresholds of the palm for modulating waves with and without a DC offset.
The results of the study show that the perceptual thresholds of different parts of the hand are different, and for different rendering methods, the perceptual thresholds of the hand are also different. This indicates that, in the human-computer interaction applications that utilize ultrasonic haptics, to render a fixed-intensity tactile stimulus, the ultrasonic intensity should be dynamically changed according to the rendering conditions. This shows the importance of considering rendering methods for ultrasonic tactile representation systems to produce a more realistic tactile experience.
This study is only a preliminary investigation into the relationship between the perceptual threshold of the hand and different ultrasound haptics rendering conditions, and future studies may explore this topic more deeply. In this study, amplitude modulation is used in ultrasound haptics rendering, whereas ultrasonic haptics based on other modulation methods can be a topic of future studies.

Reference

1.

Hayward V, Cruz-Hernandez M. Tactile display device using distributed lateral skin stretch. In: Proceedings of The Haptic Interfaces for Virtual Environment and Teleoperator Systems Symposium. New York, USA, ASME, 2000, 69(2): 1309–1314

2.

Wiesendanger M. Squeeze film air bearings using piezoelectric bending elements. Verlag nicht Ermittelbar, 2001

3.

Mallinckrodt E, Hughes A L, Sleator W. Perception by the skin of electrically induced vibrations. Science, 1953, 118(3062): 277–278 DOI:10.1126/science.118.3062.277

4.

Obrist M, Seah S A, Subramanian S. Talking about tactile experiences. In: Proceedings of the SIGCHI Conference on Human Factors in Computing Systems. Paris, France, ACM, 2013, 1659–1668 DOI:10.1145/2470654.2466220

5.

Frier W, Ablart D, Chilles J, Long B, Giordano M, Obrist M, Subramanian S. Using spatiotemporal modulation to draw tactile patterns in mid-air//Haptics: Science, Technology, and Applications. Cham: Springer International Publishing, 2018, 270–281 DOI:10.1007/978-3-319-93445-7_24

6.

Iwamoto T, Tatezono M, Shinoda H. Non-contact method for producing tactile sensation using airborne ultrasound// Haptics: Perception, Devices and Scenarios. Berlin, Heidelberg: Springer Berlin Heidelberg, 504–513 DOI:10.1007/978-3-540-69057-3_64

7.

Hoshi T, Takahashi M, Iwamoto T, Shinoda H. Noncontact tactile display based on radiation pressure of airborne ultrasound. IEEE Transactions on Haptics 2010, 3(3): 155–165 DOI:10.1109/toh.2010.4

8.

Inoue S, Makino Y, Shinoda H. Producing airborne ultrasonic 3D tactile image by time reversal field rendering. In: SICE Annual Conference (SICE). Sapporo, Japan, IEEE, 2014, 1360–1365 DOI:10.1109/sice.2014.6935269

9.

Spelmezan D, Gonzalez R M, Subramanian S. SkinHaptics: Ultrasound focused in the hand creates tactile sensations. In: Haptics Symposium (HAPTICS)．Philadelphia, PA, USA, IEEE, 2016, 98–105 DOI:10.1109/haptics.2016.7463162

10.

Jones L A, Lederman S J. Applied aspects of hand function//Human Hand Function. Oxford University Press, 2006, 179–203 DOI:10.1093/acprof:oso/9780195173154.003.0010

11.

Hasegawa K, Shinoda H. Aerial vibrotactile display based on multiunit ultrasound phased array. IEEE Transactions on Haptics 2018, 11(3): 367–377 DOI:10.1109/toh.2018.2799220

12.

Kervegant C, Raymond F, Graeff D, Castet J. Touch hologram in mid-air. In: ACM SIGGRAPH 2017 Emerging Technologies. Los Angeles, California, USA, ACM, 2017, 23 DOI:10.1145/3084822.3084824

13.

Long B, Seah S A, Carter T, Subramanian S. Rendering volumetric haptic shapes in mid-air using ultrasound. ACM Transactions on Graphics 2014, 33(6): 1–10 DOI:10.1145/2661229.2661257

14.

Linschoten M R, Harvey L O, Eller P M, Jafek B W. Fast and accurate measurement of taste and smell thresholds using a maximum-likelihood adaptive staircase procedure. Perception & Psychophysics 2001, 63(8): 1330–1347 DOI:10.3758/bf03194546