1 Introduction

Locomotion interfaces have been developed to create the bodily sensation of walking in a virtual space. These systems typically require a large physical space if voluntary walking motion of the user is allowed, and/or a device that cancels the user’s physical movement, such as an omnidirectional treadmill [1]. Otherwise, the direction of active real walking can be changed or modified to keep the user within a limited area, as in the walk-in-place [2] and the redirect walking [3] methods.

An alternative approach for inducing the sensation of walking is to move parts of the user’s body using mechanical devices. This is a form of passive body motion, in which the user does not move their body by themselves. The passive condition is appropriate when the user receives an experience of a particular virtual space that presents a first-person tour of a past real world scene, such as a multimodal movie. This may allow the user to share the body motion of another person who visited the place in the past. The simulated place may be a location of interest, such as a world heritage site, a high mountain, a beautiful beach, or a museum. Reproducing the experience of walking is a basic element of experience-sharing. We have developed a multisensory display for this purpose [4].

In the current study, we developed and tested a passive body-stimulation method to induce the sensation of walking up and down stairs, using a vestibular display (Fig. 1) while the user was sitting in real space. First, sensation scaling on the display motion was performed to show the characteristics of the stimuli presented by the display. The sensation of real walking was measured using nine factors as a reference for the design of the display. We tested participants walking on a flat floor and stairs, collecting data in each condition. Participants then adjusted the magnitude of motion on a vestibular display (a motion seat with three degrees of freedom) to create the sensation of walking on a level floor and up or down stairs. The selected magnitude of motion of the display was discussed.

Fig. 1.
figure 1

A vestibular display (motion seat) with three degrees of freedom (lift, roll, pitch motions).

Full size image

2 Sensation Scaling of Vestibular Display Stimulus

The psychological perception levels experienced by participants in response to the display motion were measured using sensory scaling in each degree of freedom (lift, roll, and pitch directions) separately.

2.1 Participants and Procedure

Participants in the experiment were five undergraduate and graduate students (average age: 22.8 years) for the lift stimulus, five students (average age: 22.4 years) for the roll stimulus, and six students (average age: 23.2 years) for the pitch stimulus.

Participants walked on a flat floor for 20 m with a walk period of 1.4 s (0.7 s each step) and remembered the sensations of the body motion of lifting, pitch and roll rotations during the walk. Participants then sat on the vestibular display and received one of the stimuli randomly selected from a set. They compared the stimulus with a standard stimulus and rated the sensation of each stimulus using a visual analogue scale (VAS). Participants closed their eyes and wore noise-emitting ear-phones during stimulation.

The standard stimulus was a 1.25 mm lift, a 0.68° roll rotation, and a 0.55° pitch rotation in a sinusoidal trajectory. The modulus was 10. The comparison stimuli of lift had seven levels of {0.02, 0.08, 0.33, 1.35, 5.50, 22.38, 91.20} mm with a lift speed of 100 mm/s rise and fall. The comparison stimuli of roll had 33 levels of degree {0.03, 0.06, 0.10, 0.19, 0.36, 0.68, 1.26, 2.34, 4.37, 8.13, 15.0} and deg/s {1.8, 4.5, 7.2}. The comparison stimuli for pitch had 36 levels of degree {0.03, 0.06, 0.13, 0.27, 0.55, 1.14, 2.34, 4.84, 10.0} and deg/s {0.7, 2.9, 5.0, 7.2}.

2.2 Result

The results (lift, roll, pitch) are shown in Figs. 2, 3, and 4. The non-linearity of sensation intensity was relatively large, with around 0.4 to 0.6 of the power index of the linear approximation on a log-log plot. The pitch sensation intensity was represented by two lines. This result may indicate that the small stimulus was perceived only by cutaneous sensation, and not by vestibular sensation.

Fig. 2.
figure 2

Sensation scale of lift motion.

Full size image
Fig. 3.
figure 3

Sensation scale of roll motion.

Full size image
Fig. 4.
figure 4

Sensation scale of pitch motion.

Full size image

3 Body Motion in Real Walking

Body motion trajectories during three real sessions of walking (walking on a level surface, walking up stairs, walking down stairs) were measured with an optical sensor (OptiTrak, V120), to be used as a reference for the motion of the vestibular display. Head motion was also measured with an acceleration sensor (TSND-121). Participants walked on a treadmill and up or down stairs (stair riser: 160 mm, tread width: 320 mm) with a walk period of 1.4 s, with markers attached at the head and the coxal bone (Fig. 5).

Fig. 5.
figure 5

Marker sites on the participant.

Full size image

The body motion during real walking was as follows:

  • Level walking: 30 to 40 mm lift, 5 to 6° roll rotation, and 1 to 2° pitch rotation, respectively. The vertical acceleration of the head was −1 to 5 m/s2.

  • Walking up stairs: 160 to 170 mm lift, 10 to 11° roll rotation, and 2 to 3° pitch rotation, respectively. The vertical acceleration of the head was −3 to 7 m/s2.

  • Walking down stairs: −160 to −180 mm lift, 3 to 4° roll rotation, and 1.5 to 3° pitch rotation, respectively. The vertical acceleration of the head was −3.5 to 7 m/s2.

The amplitude of lift motion primarily depended on the step height of the stairs. Regarding the acceleration of the head, the range of acceleration was greater for stair walking compared with walking on a level surface. The amplitude of roll and pitch motions was greater during stair walking compared with level walking. The head roll while walking up stairs was greater than that when walking down stairs, as participants shifted their body weight to the support leg and lifted the body, creating a roll motion of the head. These characteristics were necessary to reproduce with the vestibular display.

4 Evaluation of Walking Sensation During Real Walking

The sensations of level walking, walking up stairs, and walking down stairs were first analyzed to compare them with the sensations generated by the vestibular display. The dimensions of the stairs were the same as those described in the previous section.

4.1 Participants and Procedure

Participants were 13 undergraduate and graduate students with an average age of 22.9 years. Participants walked on a flat floor and stairs while wearing visual information reduction glasses with a walk period of 1.4 s, and were instructed to remember the sensation of walking. Participants then evaluated the sensation of walking in terms of the nine factors listed in Table 1 using a VAS ranging from no sensation (0) to a very definite sensation (100).

Table 1. Factors of walking sensation.
Full size table

4.2 Result

The results are shown in Fig. 6. The sensation intensity of power (acceleration) was significantly higher during stair walking compared with level walking (p < 0.01). The sensation of walking velocity exhibited a moderate range for both level and stair walking. Periodicity and lateral alteration were clearly perceived in all walking conditions. During stair walking, muscle tension was strongly perceived in the lower limbs as the body was raised and lowered by its own muscles. Continuous body motion was perceived more clearly in stair walking compared with level walking, since the amount of total body motion was greater in stair walking. In contrast, the regularity of body motion was greater in level walking, because of the small amount of perturbation of the posture. A moderate level of balance control was perceived during level walking, whereas balance control was clearly perceived during stair walking. Foot sole taction (tactile sensation) was clearly perceived in both the level and stair walking, with a stronger sensation in stair walking than level walking when the body was lifted on one foot, or when body weight was supported during sole contact on the ground during stair walking.

Fig. 6.
figure 6

The stairs the participant walked up and down.

Full size image

Overall, the sensation intensity was lower during level walking compared with stair walking, except for the regularity of body motion. This finding may have been caused by difficulty perceiving level walking itself, as level walking is often a relatively automatic movement performed without conscious control of the body. In contrast, during stair walking, the participant focused on controlling their bodies to ensure they did not miss the step, requiring a considerable amount of attention to be directed to information about their own body, as well as the environment. This process would be likely to cause higher ratings of walking sensation (Fig. 7).

Fig. 7.
figure 7

Results of sensation ratings of level walking and stair walking.

Full size image

The intensity of the virtual walking sensation induced by the multisensory display would be expected to increase if the sensation levels obtained in this experiment are achieved.

5 Optimization of Motion Stimulus

Our preliminary experiment revealed that presenting the same amount of vestibular display motion (lift, roll, pitch) as that involved in real body motion was too intense to be perceived as the sensation of walking. Although the mechanisms underlying this effect are unclear, we considered that this difference in sensation may have been related to at least two inconsistencies between the real and virtual (passive) experience of walking. First, the motor command and its efference copy [5] are not produced in the case of passive experience, so processing of internal motion model (body schema) in the brain differs from that in active motion. The difference includes sensory inhibition which occurs in active motion to reduce the intensity of the sensation of body motion. The other inconsistency is related to posture; seated and standing postures create different joint motion profiles and muscle activities that would be expected to produce different motion sensations.

As adequate stimulus intensity was not obtained from an unknown body schema, the amount of stimulation, the amplitude of motion (lift, roll, pitch) for the walking sensation were experimentally tested, using the method of adjustment.

5.1 Participants and Procedure

Participants for the experimental measurement were 10 undergraduate and graduate students (average age: 23.1 years) for level walking, 10 students (average age: 23.5 years) for walking up stairs, and 10 students (average age: 23.8 years) for walking down stairs.

Participants walked on a flat floor and stairs with a walk period of 1.4 s, and were instructed to remember the sensations of body motion of lift, pitch and roll rotations during walking. Participants then sat on the vestibular display and adjusted the amplitude of the three degrees of freedom motion (lift, roll, pitch rotation) and the speed (rising and falling) to produce the optimal sensation of walking, where a regular trajectory was fixed to a sinusoidal curve. Participants kept their eyes closed and wore noise emitting earphones during the session. They were instructed to maintain a regular seated posture in which the backbone, thighbone and shin were at right angles to each other.

5.2 Results

The optimal (equivalent) trajectories are shown in Fig. 8. All the amplitudes were less than 1/10 of the body motion during real walking. The adjusted amplitude and speed are shown in Figs. 9 and 10. There was a significant difference between level walking and stair walking in the lift amplitude (p < 0.0001, Holm’s multiple comparison). The amplitude of level walking was smaller than walking up stairs, and greater than walking down stairs. The speed and acceleration of first half of the trajectory when walking up and down stairs were greater than for level walking. A small amount of asymmetry was observed in speed and acceleration for stair walking (Fig. 10). We considered that this asymmetry represented the sensation of increasing load on the supporting leg and additional foot contact area during stair walking.

Fig. 8.
figure 8

Average trajectory for (a) level walking. (b) walking upstairs. (c) walking down stairs.

Full size image
Fig. 9.
figure 9

Amplitude of level walking and stair walking.

Full size image
Fig. 10.
figure 10

Velocity of level walking and stair walking.

Full size image

6 Presentation of Virtual Stair Walking

Vertical body motion while walking up and down stairs was presented by the lifting/lowering motion of the vestibular display in either direction. However, continuous multi-step stair walking cannot be achieved by lifting/lowering motion only, because of the limitations of the actuator stroke. Thus, it was necessary to return the seat to the start position after a limited number of steps. The characteristics of the return motion effect were investigated.

6.1 Participants and Procedure

Ten graduate and undergraduate students (average age: 23.5 years) participated in an evaluation experiment to compare four types of return motion. The second half (the return component) of the adjusted trajectory (Fig. 8) was changed in four ways; reduced to 0 (the vestibular display continued lifted or lowered), 1/3 (return ratio 1/3), 2/3 (return ratio 2/3), 1 (complete return).

First, the participant walked on a flat floor and stairs with a walk period of 1.4 s, and made an effort to remember the sensation of the body motion. The participant then sat on the vestibular display with the backbone, thighbone and shin at right angles to each other. The stimuli were presented in a random order. The participant evaluated the walking sensation using a VAS ranging from no sensation (0) to the equivalent of real walking (100). The participant closed their eyes and wore headphones emitting noise during the session.

6.2 Results

Figure 11 shows the sensation of walking for the four return ratios. While walking up stairs, a return ratio of 2/3 was the strongest sensation, and showed a significant trend (p = 0.06) compared with return ratio 1. The sensation intensity for walking down stairs was significantly different among the return ratios (p < 0.05). These results indicate that the motion of walking both up and down stairs was required to present the sensation of walking up stairs, and for walking down stairs to a lesser extent.

Fig. 11.
figure 11

Sensation of walking.

Full size image

This may reflect the profile of real stair walking. The head motion trajectory in real walking, measured in the previous experiment is shown in Figs. 11 and 12. These trajectories indicated that the head exhibited acceleration in both the up and down directions during each step when walking up or down stairs. There was downward motion of 7 mm while walking up stairs, and an upward motion of 10 mm while walking down stairs. Thus, the results suggested that motion with a return ratio of less than 1 is suitable for inducing the sensation of walking on stairs (Fig. 13).

Fig. 12.
figure 12

Vertical head motion while walking up stairs.

Full size image
Fig. 13.
figure 13

Vertical head motion while walking down stairs.

Full size image

7 Conclusion

In the current study, we tested a method for inducing the sensation of walking using a vestibular display. The adjusted trajectory of the display motion to represent walking was less than 10% of real walking motion. Regarding the presentation of stair walking, the results suggested asymmetry in the motion trajectory. For walking up stairs, a return motion ratio of less than one was appropriate for increasing the sensation.

Future studies should examine the integration of the vestibular display with displays of other modalities. Moreover, it may be useful for future research to investigate the sensation of active control of the body (agency).