//Logo Image
Author: Yeh-Liang Hsu, Chong-Fai Wang, Chun-Ming Lien, Chung-Yang Kao (2000-04-10); Modified: Yeh-Liang Hsu (2003-02-07); recommended: Yeh-Liang Hsu (2000-04-17).
Note: This paper is published in Journal of the Chinese Institute of the Industrial Engineers, Vol. 20, No. 1, January, 2003, p. 62~70.

Design and comfort evaluation of a novel locomotion device for training overhead traveling crane operators in a virtual environment

Abstract

The development of locomotion devices that provide users a sense of walking while their positions remain localized in the physical world is still one of the major problems in current virtual reality research. This paper presents the design and evaluation of a novel locomotion device for training the overhead traveling crane operators in a virtual environment. The locomotion device employs a circular platform with hundreds of spherical rollers on it. A hoop combined with an airbag is used to constrain the position of the user, so that the user can walk on the rollers in a “step-and-slide” pattern. The user can also walk and change directions freely without holding on the hoop. When the user steps on the rollers, the switches underneath the rollers sense the motion of the feet. Two-dimensional walking data can be acquired successfully, and the user does not have to wear any special shoes or obstructive sensors other than the head-mounted display. A subjective comfort evaluation is also performed to compare the sense of walking on the locomotion device and that of natural walking.

Keywords: virtual reality, operator training system, locomotion.

Introduction

The development of virtual reality (VR) and virtual environment (VE) technologies has attracted considerable interests from industrial engineers. Wilson [1999] pointed out that VR/VE has the potential as a tool to support many types of ergonomics contributions, including assessments of office and workplace layouts, testing consequences for reach and access, reconfiguring and evaluating of alternative interface designs, checking the operating or emergency procedures, and training for industrial and commercial tasks.

In particular, virtual environments for training have shown to be a good addition or alternative to the traditional operator training systems. Tam et al. [1998] developed a PC-based VR operator training simulator for power-utility personnel. The same research group also developed a VR-based learn-by-doing system that is accessible from a Web browser for the training of maintenance workers in the troubleshooting and inspection of power transformers [Tam et al., 1999]. Arroyo et al. [1999] presented a low-cost, PC-based, virtual reality application to train personnel associated with the manual operation of electrical substations equipments. Matsubara et al. [1997] built an operator training system for training the control activities of an electric power plant using the conception of the virtual learning environment. Wilson et al. [1998] developed a virtual environment combining 3-D factory drawings with 3-D simulation models of crane and load motion for training overhead crane operators. Virtual environments are ideal for explorative and self-paced training. Such applications for operator training can provide real time interaction with virtual devices and the function to change viewpoint freely, and they also allow employees to train without taking production equipment out of service.

The function of “locomotion,” to allow the users to move about the virtual environment, is required for many training systems. Many VR/VE applications allow the users to move in a large virtual space in a vehicle. Using the control devices on the vehicle such as wheels, joysticks, buttons, pedals, etc., the user interacts with and moves in the virtual space displayed in the head-mounted display (HMD), while the position of the user (and the vehicle) is fixed.

However, the most intuitive way to move about the real world is to travel on foot. Some VR/VE applications allow the users to walk in a confined area. The three-dimensional magnetic sensors attached to the user are often used for position sensing when the user moves, so that the image of the virtual space displayed in the HMD changes accordingly. But in this approach, the sensing range of the sensors and the size of the confined area strictly limit the movable area of the user. Therefore this approach is not suitable for moving in a large virtual space. As Iwata [1999] pointed out, the development of locomotion devices that provide users a sense of walking while their positions remain localized in the physical world is still one of the major problems in current virtual reality research.

One common approach for virtual walking is to use a treadmill. Treadmills for physical fitness are unidirectional. Therefore the key issue of this approach is how to redesign the treadmill so that users are free to walk in any direction. Darken et al. [1997] developed an “omni-directional treadmill” that employs two perpendicular treadmills and allows its users to walk or jog in any direction of travel. For similar purposes, Noma and Miyasato [1998] designed a locomotion interface using a treadmill mounted on a three-axis motion platform. Recently Iwata [1999] presented the “torus treadmill.” In this system, 12 treadmills are connected side-by-side, on which users can walk in any direction.

The mechanisms of the locomotion devices using the treadmill approach are usually quite complicated. Iwata and Fujii [1996] developed the “virtual perambulator” using a much simpler “sliding surface” approach, as shown in Figure 1. In the virtual perambulator, a user wears omni-directional roller-skates or low friction rubber sandals to slide on the floor. A hoop is set around the user’s waist by which the position of the walker is limited, while the user can walk in any direction. Novice users can hold the hoop to keep their balance, and trained users can push their waists against the hoop while walking. Magnetic sensors measure the motion of the feet of the user, and the image of the virtual space displayed in the HMD refreshes accordingly. Two hundred and thirty five novice users tested this system, and 94% of the participants could freely walk around the virtual space.

Figure 1. The locomotion device using the “sliding surface” approach

This paper describes a system for training the overhead traveling crane operators in a virtual environment. The design and evaluation of the locomotion device in this training system is emphasized. The locomotion device also uses the “sliding surface” approach. But instead of wearing special shoes, the locomotion device employs a circular platform consisting of 9 plates. There are 256 spherical rollers in a  array on each plate to generate the “sliding effect.” The user can walk on the rollers in a “step-and-slide” pattern. When the user steps on the rollers, the switches underneath the rollers sense the motion of the feet. A hoop combined with an airbag is used to constrain the position of the user. The airbag also supports the waist of the user so that the user can walk freely without holding on the hoop. In this system, the user does not have to wear any special shoes or obstructive sensors. The only device the user needs to wear is the HMD.

The design requirements followed by the design of the locomotion device for training overhead traveling crane operators are described in this paper. Then a comfort evaluation is also performed to compare the sense of walking on the locomotion device and that of natural walking, and conclusions are drawn.

Training system for overhead traveling crane operators

The overhead traveling crane is widely used in manufacturing factories to transport materials and loads from one point to the other. The main structure of an overhead traveling crane is an I-beam that is driven by an electric motor to travel on a fixed track. Another motor controls the hoisting mechanism for lifting loads. The operator holds a crane controller with both hands to hoist up loads, and to drive the crane to move on the track. Usually the operator walks with the crane as it travels.

Several studies concerning the structural safety, e.g., the fatigue strength of the structure, of the overhead traveling crane were carried out [Kitsunai et al., 1998; Zai et al., 1994]. However, a study of local factories in Taiwan shows that, most accidents related to the overhead traveling crane are not caused by structural failure. Instead, the major cause of accidents is that the crane is not properly operated. Operating the crane, to control simultaneous traveling, traversing, and hoisting motions of the crane and the resulting load swing, requires a lot of skills. The crane has to be carefully controlled and moved at low speeds to avoid dangerous load oscillations. The complicated dynamic motion of the overhead traveling crane also attracts the attention of many researchers in automatic control [Al-Garni et al., 1995; Lee, 1998]. However, currently overhead traveling cranes are still controlled by human operators. Therefore, training of the overhead traveling crane operators is very important.

Traditionally, training of overhead traveling crane operators often takes place in a real factory where the trainee practices operating a real crane. The VR/VE technology provides an ideal solution to the training system for the overhead traveling crane operators. The trainee can practice safely in a virtual environment. Different factory setups and operating scenarios can be programmed in the training system to give the trainee more experience before operating a real crane in the factory. In order to let the user walk in a large “virtual factory,” a locomotion device becomes essential. The major design requirements for the locomotion device of the training system of overhead traveling crane operators are:

1.      The locomotion device should allow free, natural two-dimensional walking, while the position of the walker remains localized.

2.      To provide the sense of “reality,” the motion of the feet should be detected and transferred to the VR software rapidly, so that the image of the virtual factory displayed in the HMD can refresh in real time.

3.      Since the user has to operate the crane controller with both hands, the user should be able to walk safely in the locomotion device without holding anything with hands.

Design of the locomotion device

Figure 2 shows our design of the locomotion device. There are three major components: the main frame (component 1 in Figure 2), the roller plate (component 2), and the waist support (components 3, 4, and 5).

Figure 2. The locomotion device

We used the “sliding surface” approach in this design. To achieve free and natural two-dimensional walking, we first decided that the user does not have to wear any special shoes or obstructive sensors other than the HMD. As shown in Figure 3(a), Fruin [1971] defined the sizes of buffer zone areas, including “touch zone (A),” “no-touch zone (B),” “personal zone (C),” and “circulation zone (D).” The locomotion device employs a circular platform of the same size as the “circulation zone,” within which circulation is possible without disturbing others. The circular platform consists of 9 plates, with 256 rollers in a  array on each plate, as shown in Figure 3(b). The rollers are used to generate the “sliding effect.” The user walks in a “step-and-slide” pattern on these rollers.

 

Figure 3(a) Buffer zones [Fruin, 1971]                  Figure 3(b) The roller plate          

Similar to the virtual perambulator mentioned in the previous section, the user’s waist is also supported by a hoop while walking on the platform. The height of the hoop can be adjusted. An airbag attached to a ring at the inner surface of the hoop is inflated to pressure the waist of the user, so that the user does not have to intentionally push the waist against the hoop. This design enables the user to walk naturally and safely without holding anything with the hands. There are roller bearings between the hoop and the ring that the airbag is attached. The user can twist the waist freely while the airbag tightly presses the waist. Note that the supporting plate of the hoop inclines 45 degrees downward, so that the user can swing both arms freely while walking on the device.

The rollers on the plates are also used as sensors for the motion of the user’s feet. Figure 4 shows the configuration of a roller. A steel ball (0.75 in. in diameter) is supported by 6 small steel balls (0.25 in. each in diameter). A normally closed (N/C) switch is mounted under the steel balls. As shown in Figure 4, when the user steps on the balls, the spring is compressed and the switch opens.

                 

Figure 4. The configuration of a roller

Figure 5 shows how the motion of feet is detected using these switches. Switches on each plate are arranged into a sequential array circuit (Figure 6), and are scanned iteratively. Signals obtained from the 9 plates are mapped into a large array after filtering the noisy signals. Finally, positions of both feet at a given time are calculated and output to the VR software for further processing. Combining with signals from the tracker of the HMD, the speed and direction of the user are computed, the VR software refreshes the image displayed in the HMD accordingly.

Figure 5. Sensing the motion of the feet

Figure 6. The sequential array circuit

Figure 7 is the walking data collected when the user walks in one direction. The peaks represent the positions of the feet. The horizontal axis is the time elapsed. Obviously, the positions of the user’s left and right foot change alternatively. This data is used to compute the velocity of the user. Note that the user follows a “step-and-slide” walking pattern when walking on the locomotion device. The switches on the platform return signals only during the “slide” half of the walking pattern. Therefore, the left half of each peak in Figure 7 is a sudden and discontinuous change. To avoid generating an abrupt change of velocity, a linear interpolation is used to estimate the position of the user’s feet during the “step” half of the walking pattern.

Figure 7. Data for unidirectional walking

The update rate of the image displayed in the HMD, or often called the “frame rate,” is very crucial to the sense of presence of the user. In order to achieve a sufficiently high frame rate, only a quarter of the switches are scanned, as indicated by the solid circles in Figure 3(b). With this resolution, the smallest movement of the feet that can be detected is about 50 mm.

Figure 8 shows the positions of both the feet on the x- and y-axis when the user walks and gradually turns right. Accumulating the walking distances of each step, a trajectory of the user is formed in the virtual environment as in Figure 9. These figures show that two-dimensional walking data can be acquired successfully.

(a) Positions of both the feet on the x-axis

(b) Positions of both the feet on the y-axis

Figure 8. Data for two-dimensional walking

Figure 9. Trajectory of the user in two-dimensional walking

Finally, VR scenes for overhead traveling crane operator training are constructed. The VR software used in this research is “World Up” by Sense 8 Corporation (http://www.sense8.com/), which provides real-time functionality in an interactive, object-oriented environment. The HMD used in this research is “I-glasses X2” by I-O Display Systems (http://www.i-glasses.com/). The HMD has 2 full-color 0.7” LCDs, 360,000 pixels per eye in standard mode, and 180,000 pixels per eye in 3D, which has the capability of true stereoscopic imaging. The HMD weighs about 220g, and has fully adjustable stereo earphones.

Figure 10 shows the photo of a complete training system. The user first stands inside the waist support hoop, then the hoop is lifted to a proper location and the airbag is inflated. The user wears an HMD, and holds an overhead crane controller with both the hands. Through the controller, the user interacts with the overhead traveling crane in the VR scenes, to hoist loads and drive the crane to move on the track. The user can also walk in any direction as the crane travels.

Figure 10. A photo of the complete system

Subjective comfort evaluation of the new locomotion device

As mentioned earlier, VR training systems receive more and more attention for operator training. However, there are limitations to current VR training systems, and the operating dynamics of a VR working platform will not exactly be the same as that of the real world. As discussed in previous sections, whether the locomotion device can provide a natural sense of walking, that is, whether the user can walk comfortably on the device, is really the key to the success of the overhead traveling crane operator training system.

The comfortable level evaluation of a device falls on the observations of body postures and movements, observations of task performances, and direct subjective ratings of general comfort. The main action of our locomotion device is “walking.” Since the change of body postures and the task performances while walking are not easy to observe and evaluate, a subjective comfort evaluation was performed to compare the sense of walking on the new locomotion device with that of natural walking.

Eighteen healthy subjects, fourteen males and four females, volunteered to participate in this study. The averaged age was 23.6 years, and they all had normal walking ability. Neither did the subjects participate in the development of this locomotion device, nor did they have any knowledge about the locomotion device before the test. Before the first measurement, each subject was informed about the experimental procedure and agreed to answer the questions sincerely.

Each subject experienced three different walking tests: (1) natural walking on the ground; (2) walking on the locomotion device without any practice; and (3) walking on the locomotion device with some practice. In the first two tests, the facilitator guided the testers to walk on the locomotion device or normal ground by choosing them randomly, without telling the testers what surfaces they are walking on. The third test was performed after all testers finished the first two tests. Each walking test lasted for about 3 minutes. Guided by the fascinator, the testers were asked to walk in one direction for 1 minute, then to walk freely in any directions for 2 minutes.

Ideally the subjects should wear the HMD while walking on the locomotion device. But the scenes displayed in the HMD are controlled by the locomotion device. It is not feasible to wear HMD and do natural walking on the ground. Furthermore, the difference between the VR scenes and the surroundings of the real world may further affect the subjects’ evaluation of the sense of walking on the locomotion device. Therefore, the subjects’ eyes are covered in all three tests to avoid the effect from the perception of the eyes, and they were told to concentrate on the feeling of all parts of the body while walking. The evaluation here is strictly on whether the locomotion device can provide a natural, comfortable feeling of walking.

The purpose of the second test was to investigate users’ response when they use the locomotion device for the first time. Before the third test, the subjects were specifically told to practice the “step-and-slide” walking pattern for a few steps. The purpose of the third test was to investigate whether the users would adapt themselves to the locomotion device after some practice.

The subjects received a 10 minute break between each test. During the break, the subjects were asked to answer 6 questions subjectively on an eleven-point scale (‘5’ being ‘very comfortable’, ‘0’ being ‘adequate’, and ‘-5’ being ‘very uncomfortable’), and to describe their experience about the test. The 6 questions were, while walking, whether the subjects were comfortable (1) around the shoulder; (2) around the waist; (3) twisting the waist; (4) swinging both arms; (5) around the legs; and (6) whether the subjects were comfortable with their walking patterns.

Figure 11 shows the average scores of each question rated by the eighteen testers. While walking on the locomotion device, the testers felt comfortable around the shoulder, twisting the waist, swing the arms, and around the legs, though the scores of the last three areas mentioned above are significantly lower than the scores of natural walking.

Figure 11. The result of the human description evaluation

The testers gave only an “adequate” rating to the pressure around the waist in tests (2) and (3). As discussed in the previous section, there is an airbag around the inner surface of the waist hoop. The airbag is inflated to pressure the waist of the user tightly, so that the user does not have to hold anything with the hands or intentionally push against the hoop. However, this pressure around the waist also gave a strong sense of being constrained, which does not exist in natural walking.

In Iwata and Fujii’s [1996] “virtual perambulator” discussed in the first section, 94% out of the 235 novice users could freely walk around the virtual space. Here all testers could walk freely around the virtual space, but when asked whether they are comfortable with the “walking pattern” in the questionnaire, almost all testers gave “less-than-adequate” ratings. Obviously, the sense of the “step-and-slide” walking pattern is different from that of natural walking. The users need some practice to get used to this walking pattern, just like practice to get used to walking on a treadmill. From the results of the second and third walking tests shown in Figure 11, the testers’ response on walking pattern did improve after some practice.

The results of ANOVA analysis in Table 1 shows that the overall comfort rating of “natural walking on the ground” is significantly higher than those of “walking on the locomotion device without practice” and “walking on the locomotion device with practice”P<0.05. The difference of overall comfort rating between “walking on the locomotion device without practice” and “walking on the locomotion device with practice” is not statistically significant.

Table 1. Results of ANOVA analysis

Walking pattern

F

P

TLV

Natural walking on the ground & Walking on the locomotion device without practice

8.610

0.015

4.965

Natural walking on the ground & Walking on the locomotion device with practice

8.724

0.014

4.965

Walking on the locomotion device without practice & Walking on the locomotion device with practice

0.023

0.882

4.965

 

 

 

α=0.05

Conclusions

The development of virtual reality and virtual environment technologies provides a powerful tool for assessments of working environments. Especially in training, VR/VE devices have the advantage of let the users to explore and practice situations or behaviors that are dangerous or unavailable using “real devices.” In the visual aspect, many researches in VR focus on the visual aspect of how to generate 3D models that the user can interact with intuitively in real time. In the mechanical aspect, whether VR/VE devices can provide a natural, comfortable sense of body movement as in the “real devices,” is also very important.

The development of locomotion devices that provide users a sense of walking while their positions remain localized in the physical world is still one of the major problems in current virtual reality research. This paper presents the design and comfort evaluation of a novel locomotion device for training the overhead traveling crane operators in a virtual environment. The locomotion device employs a circular platform with hundreds of spherical rollers on it. A hoop combined with an airbag is used to constrain the position of the user, and the user can walk on the rollers in a “step-and-slide” pattern. The airbag supports the waist of the user so that the user can walk and change directions freely without holding on the hoop. When the user steps on the rollers, the switches underneath the rollers sense the motion of the feet. Two-dimensional walking data can be acquired successfully, and the user does not have to wear any special shoes or obstructive sensors other than the HMD.

A comfort evaluation is performed to compare the sense of walking on the locomotion device and that of natural walking. The testers felt comfortable with the locomotion device in general, and they are able to walk freely on the device. Though the testers are not used to the “step-and-slide” walking pattern, their response on the walking pattern improved after some practice. The testers also responded that the waist support of the device was too rigid, which gave them a strong sense of being constrained.

With these feedbacks from the testers, some suggestions are made for future development. Some kind of cushion should be added between the inner surface of the waist hoop and the airbag ring. This should soften the constrained feeling of the users and allow the users to lean forward while walking. Note that the surface of the unidirectional treadmill for physical fitness always inclines backward, which creates a “climbing effect” that makes the user’s slide motion more natural. Analogous to the treadmills, another possible improvement for future development of the locomotion device is to change the roller plate into a bowl-type surface. This is technically much more difficult but should improve the sense of the “step-and-slide” walking pattern on the locomotion device.

Acknowledgement

Financial support from the Institute of Occupational Safety and Health, Taiwan, ROC is gratefully acknowledged.

Reference

Al-Garni, A.Z., Moustafa, K.A.F., Javeed Nizami, S.S.A.K., 1995. “Optimal control of overhead cranes,” Control Engineering Practice, v 3, p 1277-1284.

Arroyo, E., and Los Arcos, J.L., 1999. “SRV: a virtual reality application to electrical substations operation training,” International Conference on Multimedia Computing and Systems -Proceedings, v 1, p 835-839.

Darken, R.P., Cockayne, W.R., and Carmein, D., 1997. “Omni-directional treadmill: A locomotion device for virtual worlds,” Proceedings of the ACM Symposium, p 213-221.

Iwata, H., and Fujii, T., 1996. “Virtual Perambulator: A novel interface device for locomotion in virtual environment,” Proceedings - Virtual Reality Annual International Symposium, IEEE, p 60-65.

Iwata, H., 1999. “Torus Treadmill: Realizing locomotion in VEs,” IEEE Computer Graphics and Applications, v 19, p 30-35.

Fruin, J.J., 1971. Pedestrian Planning and Design, New York: Metropolitan Association of Urban Designers and Environmental Planners, Inc., pp.64-69

Kitsunai, Y., Maeda, Y., and Yoshihisa, E., 1998. “Fatigue strength of gusset welded joints under simulated program loading of overhead travelling crane,” JSME International Journal, Series A, v 41, p 112-120.

Lee, H.-H., 1998. “Modeling and control of a three-dimensional overhead crane,” Journal of Dynamic Systems, Measurement and Control, Transactions of the ASME, v 120, p 471-476.

Matsubara, Y., Toihara, S., Tsukinari, Y., and Nagamachi, M., 1997. “Virtual learning environment for discovery learning and its application on operator training,” IEICE Transactions on Information and Systems, v E80-D, p 176-188.

Noma, H., and Miyasato, T., 1998. “Design for locomotion interface in a large scale virtual environment ATLAS: ATR locomotion interface for active self motion,” American Society of Mechanical Engineers, Dynamic Systems and Control Division (Publication), DSC v 64, p 111-118.

Tam, E.K., Maurel, C., Desbiens, P., Marceau, R.J., Malowany, A.S., and Granger, L., 1998. “Low-cost PC-oriented virtual environment for operator training,” IEEE Transactions on Power Systems, v 13, p 829-835.

Tam, E.K., Badra, F., Marceau, R.J., Marin, M.A., and Malowany, A.S., 1999. “Web-based virtual environment for operator training,” IEEE Transactions on Power Systems, v 14, p 802-808.

Wilson, B., Mourant, R.R., Li, M., and Xu, W., 1998. “Virtual environment for training overhead crane operators: Real-time implementation,” IIE Transactions (Institute of Industrial Engineers), v 30, p 589-595.

Wilson, J.R., 1999. “Virtual environments applications and applied ergonomics,” Applied Ergonomics, v 30, p 3-9.

Zai, J., Cao, J., and Bell, A.J., 1994. “Fatigue strength of box girders in overhead travelling cranes,” Structural Engineer, v 72, p 401-405.