Applied Computer Simulation Labs

  • Increase font size
  • Default font size
  • Decrease font size

Virtual Reality Solutions for Children with Physical Disabilities

Virtual Reality Solutions for Children

With Physical Disabilities

The Second International Conference

on the Military Applications of Synthetic

Environments and Virtual Reality


Dr. Dean Inman (presenter), Mr. Ken Loge, and Mr. John Leavens

Oregon Research Institute

Applied Computer Simulation Labs

Eugene, Oregon, U.S.A.


The notion that childhood development is directly related to being able to independently explore one's environment is now widely accepted among social scientists, cognitive psychologists, and early childhood education specialists. A significant body of research has presented evidence over the last twenty-five years that self-locomotion experience plays an important role in the development of spatial perceptual abilities and cognition (Held and Hein, 1963; Bertenthal, Campos and Barrett, 1984; Adolph, Gibson and Eppler, 1990; Acredolo and Evans, 1980; Kermoian and Campos, 1988; Bertenthal and Bai, 1987; Horobin and Acredolo, 1986). These skills form an integral part of how we interact with and utilize our environment.

Spatial perception encompasses our ability to localize ourselves in space, depth perception, shape recognition, visually guided reaching, awareness of self-motion, as well as mapping and problem solving in multidimensional space. This set of skills is necessary for many of the activities of daily life including mobility, vocational tasks, and independent social interaction. Children with motor dysfunctions often do not have access to self-locomotion experience during early developmental years (birth to five years). For children whose motor dysfunction is severe, mobility experience can be delayed until young adulthood and beyond. The effects of this deficit in ability experience is severe.

Fortunately, adaptive equipment is available to compensate for many physical limitations. When mobility is limited a motorized wheelchair is an excellent means by which children can achieve independent mobility. However, these wheelchairs are very expensive and it is unlikely that a motorized wheelchair will be purchased for a child unless it is clear in advance that (s)he will be able to use it effectively. Unfortunately, proficient use may require substantial training. This results in a "Catch-22" situation in which a child who needs a wheelchair with which to practice in order to acquire the necessary skills cannot obtain the wheelchair until a satisfactory level of proficiency has been achieved.

Children who fail the initial evaluation are customarily denied access to motorized wheelchairs, and thus never experience the increased mobility that would have been afforded to them by a motorized wheelchair. Even children who already own motorized wheelchairs frequently lack the skills required to operate them safely in all environments. Enhancing their skill level requires intensive one-on-one training and close supervision to prevent mishaps and avoid injury.

What is needed is a cost-effective means of training children to criterion without having to purchase motorized wheelchairs and/or without exposing them and others to physical danger during the training process. Virtual reality, a newly emerging technology, is exciting in its implication for providing such training. This technology permits an individual to experience apparent movement and to explore lifelike environments. Its advantages for teaching complex motor skills are well-documented. For example, virtual reality has been used in the training of airline pilots, astronauts and surgeons. It is a safe, cost-effective, means of training sophisticated perceptual-motor skills. The purpose of this project is to demonstrate and evaluate three virtual reality training programs designed to teach orthopedically challenged children to operate a motorized wheelchair safely in the natural environment.


Children included in this study are between three and fifteen years of age. They all have severe orthopedic impairments which require them to use wheelchairs for mobility. The children we see can be divided into two groups. The first group consists of children who do not have their own motorized wheelchair because they lack sufficient skills to warrant the expenditure. These children want to acquire sufficient driving skills so they can justify purchasing their own wheelchairs in the future. The second group of children already have motorized wheelchairs but they are unable to operate them safely in all environments. These children wish to improve their driving skill so they can enjoy greater independence in the community and home settings.

Training is accomplished by having children experience a series of three virtual training scenarios (Figure 1). These virtual worlds are sequenced in terms of difficulty and complexity (Figures 2, 3, and 4). All training is conducted using actual motorized wheelchairs with actual joysticks in order to facilitate generalization of skills learned in virtual reality to actual reality. The length of training sessions is determined by the student driver as is the amount of time students choose to spend in each of the three training scenarios.


The training platform is designed to accommodate actual motorized wheelchairs of various sizes. Children who already have their own wheelchairs use them during the training, while children who do not already own wheelchairs are provided one for use during training and subsequent evaluation. Several features have been incorporated into the workstations that are designed to mimic reality and thus enhance the transfer of skills to real-world settings. For example, a student's motorized wheelchair is placed on rollers supported by a metal frame approximately six inches off the ground. A small ramp enables the chair to be pushed up onto the platform and into position.

Picture of a child using the VR wheelchair training platform.

Figure 1: The virtual-reality based training environment.
(Photograph copyright Peter I. Chapman, printed with permission)

When in position, the back wheels of the wheelchair rest on rollers that permit the wheels to rotate at normal speeds but which effectively prevent the chair from actually moving. Moreover, each back tire is situated on its own set of rollers which allows the right and left wheel to move independently of the other. Thus, when the joystick is moved to the left, which would normally cause the right rear wheel to turn faster than the left wheel, the apparatus permits differential axle speed to simulate the tactile, kinesthetic, and auditory feedback normally associated with making a left hand turn. Also, the roller bearings are slightly off-center which causes the chair to vibrate very slightly as it would normally do on surfaces such as sidewalks and low-pile carpeting.

Picture of world one virtual wheelchair training environment.

Figure 2: The virtual world with low-level complexity.

Picture of world two virtual wheelchair training environment.

Figure 3: The Virtual world with intermediate-level complexity.
(Photograph copyright Peter I. Chapman, printed with permission).

Picture of world three virtual wheelchair training environment.

Figure 4: The virtual world with high-level complexity.
(Photograph copyright Peter I. Chapman, printed with permission).

Encoders are built into the roller assembly which enables the computer to determine how fast the wheelchair is moving and which direction it turns. This is accomplished by a simple bar code reader which is mounted on the roller assemblies supporting the right and left wheels of the wheelchairs.

As the roller assemblies rotate, an electric eye reads the relative speed and direction each roller is turning. This way the computer can determine if the chair is moving forward or backward, turning right or left, and how fast the chair is moving through the virtual world. Inertial aspects of the wheelchair were simulated by weighting the roller assemblies with buckshot.

Once the wheelchair is in place, the student driver is fitted with a head mounted display, through which the visual and auditory components of the virtual reality training program are presented. Inside the visor of the helmet are two video screens, one for each eye, through which a three-dimensional representation of a virtual reality world can be seen. Also in the helmet are two earphones through which a stereophonic program can be presented to coincide with the visual images displayed in the visor.

It should be noted, however, that the earphones in the helmet do not block out sound from the real world during the training process. This makes it possible to talk with students as they experience the Virtual Reality training scenarios. It also enables them to hear the motors of the wheelchair turning as they drive the rear wheels at different speeds.

Mounted on top of the head-mounted display is a head tracker which enables the computer to monitor changes in head position which, in turn, permits the software to create images of the virtual world accordingly. So, while a user is driving s/he can look right, left, up, and down and see what would be expected to be seen when driving in actual reality.


Scenarios One and Two are designed to promote independent exploration, discovery, cause and effect relationships, and visual memory (skills prerequisite to independent mobility that orthopedically-impaired children often lack). These virtual realities are safe, highly entertaining situations where the student learner is allowed to roam freely during the discovery process. Scenario Three provides a more structured environment in the community and attempts to establish appropriate street crossing skills using a crosswalk protected by a traffic light. Although many other scenarios are possible, we have limited the scope of our initial efforts to demonstrating and evaluating these three scenarios.

1. Scenario One: Exploration of an Open Environment: In the first Virtual World, the child is able to explore a large, wall-less space that has no obstacles or impediments of any kind (Figure 2). The child is free to go forward and backward at any speed within the limits of the software and hardware. When going full speed the children are able to drive approximately 80 mph (120 km). In addition, the child can choose to go in circles; large slow ones or very short tight ones. The floor consists of black and white tiles over which the child drives. This enhances the sensation of movement and speed as the child drives across the floor. In this Virtual World a child can experience the joy of independent mobility without fear and without constraints. It provides the basis upon which all other mobility skills are based.

2. Scenario Two: Exploration and Discovery in a Closed Environment: In the second Virtual World the student learner is allowed to explore an area which is approximately 2000 feet by 2000 feet. In this world, there are obstacles which the wheelchair cannot drive through (Figure 3). These obstacles must be avoided or crashes will occur. If the driver runs into anything, the chair will stop and an appropriate crash sound is heard. In order to continue driving, the chair must be backed up and then turned right or left to avoid further collisions.

Scattered around this virtual world are various items or stations that each child can discover while exploring the environment. Some of these items are obelisks, up to 50 feet high, while others are tunnel-like and drivers are able to drive through the middle of them. These stations become active and/or make sound when approached. These items are intended to be entertaining and create a sense of surprise and wonderment in the children as they independently initiate exploration in this strange and interesting world. Also included in this virtual world are virtual ice and virtual mud. Initially, these places were intended to present drivers with something to avoid while driving. On the ice the wheelchair spins out of control and driving is very difficult. In the virtual mud the chair becomes bogged down and it is difficult to get from one side of the mud puddle to another. (As it turns out however, children love finding and driving into the mud and sliding on the ice. I am sure there must be some genetic predisposition to enjoying these kinds of activities among children) Another feature of this virtual world which inspires curiosity and wonderment is that its edges are finite. In other words, it is possible to drive off the edge of the world and fly. This too is a feature that is intended to encourage children to initiate independent action and to be rewarded for the attempt.

Two other facets of Scenario Two should be mentioned. First, the student driver always starts from the same position within the world. This way, the child learns that the exploration activities always begin from the same point of departure, permitting him or her to eventually memorize the world's configuration. Second, the items or stations placed throughout the world remain constant over time. Again, this permits children to create a visual memory of the world's layout and permits them to begin exploring from where they left off at the end of the previous session or to revisit sections of the world they found particularly amusing or interesting. This consistency enhances each child's ability to (a) memorize the world's features, (b) foster systematic and deliberate exploration of an environment, and (c) discover cause-and-effect relationships between volitional activity and activation or deactivation of devices or apparatuses that are contained within the environment. This type of simulated adventure is intended to foster a sense of curiosity, personal control, and an appreciation for the benefits of independent mobility. These subjective experiences are essential to a child's ability to benefit from more advanced mobility training because they form the motivation that drives the desire to learn mobility skills.

3. Scenario Three: Initial Driver's Education for the Community: Scenario Three consists of a street crossing scenario in the community (Figure 4). Obviously, the community environment is extremely complicated with a myriad of situations that could be targeted for training. We have chosen, however, to focus on developing and evaluating a street-crossing scenario because of its frequency and importance for safe mobility in the community. This virtual reality program begins with the child on a sidewalk bordered by a lawn on one side and a two-way street with moving cars on the other. At a point 85 yards beyond the starting position, a crosswalk (complete with stoplight and pedestrian crossing sign) is available. On the other side of the street is another sidewalk perpendicular to the first one. The child's task is to (a) approach the crosswalk appropriately, (b) press the street crossing button, (c) wait for the light and crosswalk sign to operate, signaling the cars to stop, (d) drive across the street within the boundaries of the crosswalk, and (e) negotiate the turn onto the opposite sidewalk. The child is able to go back and forth across the street as many times as desired.


To date we have worked with 23 children. We now have two centers where training is being provided. One is in Portland, Oregon in a regional training program for orthopedically challenged students. The other is in Minneapolis St. Paul, Minnesota at the Sister Kenny Institute. We expect by the end of the project, July 1,1996, to have collected data on approximately 60 children. The data to be presented in this paper are not complete because they are still being collected. Therefore this section is titled Preliminary Results and data will be presented on only three children who are representative of our population sample.


The children are evaluated in actual reality. Prior to training in the virtual worlds and, subsequently, after every two hours of training time, driving skills are probed. Figure 5 shows the data which are collected. Briefly, we take data on the extent to which children are able to turn left, turn right, their ability to stop before hitting a wall and the distance they are able to travel forward along a sidewalk without going out of bounds. Three trials are taken on each driving skill. The matrix in the lower right section of the data form allows us to index the percentage of a right or left turn a driver is able to make successfully.

During evaluation the wheelchair is placed on an actual grid in a hallway and measurements are taken as to how far the driver is able to move the wheelchair into the second grid without going out of bounds. An additional dependent measure is taken when children are able to bring their own motorized wheelchairs to the training setting. Specifically, we install an odometer on the wheelchair with which to index the extent to which children use their chair between training sessions at the Oregon Research Institute.

Baseline evaluation form.

Figure 5: The Baseline Evaluation Chart.


Bert, the child involved in the study arbitrarily called Case Study One is a three-year-old child with arthrogryposis. This condition is characterized by multiple congenital contracture of the limbs. Bert is a young boy with above normal intelligence and normal vision. His upper extremities, while stiff and contractured, are well-coordinated. Our goal was to teach Bert to drive a wheelchair so he could justify getting one of his own.

With Bert's intelligence and coordination, there is little doubt that he would eventually have a motorized wheelchair of his own and would learn to drive it skillfully. The only question was when he would be able to learn the necessary skills. Our purpose in entering him into the virtual reality training project was to teach him driving skills as early as possible. We were also interested in finding out if skills learned in virtual reality would transfer or generalize to the real world.

Figure 6 shows how much total time Bert spent training in nine training sessions distributed over approximately five months. In Bert's first virtual training experience, he spent approximately three minutes training in virtual reality before electing to stop the session. The data indicate that over time Bert's willingness to work increased and towards the end of the five month interval he was spending approximately thirty minutes driving his chair in one or more of the virtual worlds.

graph of daily training time for Bert.

Figure 6: The Daily Training Time for Bert.

Figure 7 shows how Bert distributed his time among the three virtual reality training scenarios for each of the training sessions. Generally speaking, Bert's interest in working in the third virtual world increased over time and his driving skills (see below) increased accordingly. In sessions two, three and four Bert's interest in the first virtual reality training scenario increases but then he appears to have lost interest and time spent in this world falls back to levels approximating his first exposure level during Session Two. Bert's interest in exploring the second virtual world appears fairly constant across the nine training sessions.

Graph of daily training time per world for Bert.

Figure 7: Daily Training Time per World for Bert.

Figure 8 shows changes in driving skill for Bert over three probes. The data indicate that Bert was quite skillful at being able to stop before hitting a wall before he ever began training in virtual reality. His skill level remained stable over the three probes. However, Bert's ability to turn right and left and to go forward down a sidewalk without going out of bounds increased as a function of time spent training in virtual reality. Following the third and last probe Bert was graduated from our training program and he is now planning to get his own wheelchair.


John, the child involved in Case Study Two, is a four-year-old boy with multiple disabilities including cerebral palsy and mild-to-moderate retardation. He has no expressive language skills which means he is unable to communicate using speech during the training sessions. His vision is impaired but it is not known how severely, because of his inability to communicate during his eye examination.

Figure 9 shows daily training time over nine sessions. There are no obvious trends in terms of John's ability or willingness to work towards developing driving skills in virtual reality. Figure 10 shows how John distributed his time during each training session among the three virtual reality training scenarios. The figure shows that John's interest in the first virtual world appears to have decreased over time while his interest in working and exploring in the second virtual world increased. John only worked in the street crossing program on one occasion for approximately three minutes.

Figure 11 shows John's performance on two probes conducted approximately two months apart. We were able only to index his actual driving skills twice, due to his age and mental immaturity. The data show significant gains in driving ability during the second probe. Right and left hand turns were markedly improved as was his ability to drive down a sidewalk without going out of bounds. His ability to stop before hitting a wall was not indexed during the second probe period due to John's unwillingness to try this maneuver. John is still working with us to improve his driving skills.

Baseline evaluation graph for Bert.

Figure 8: Baseline Evaluation Data for Bert.

Graph of daily training time for John.

Figure 9: Daily Training Time for John.

Graph of daily training time per world for John.

Figure 10: Daily Training Period per World for John.

Graph of John's evaluation probes.

Figure 11: Evaluation Probes for John.


Paul, the child involved in Case Study Three, is a fifteen-year-old adolescent with severe cerebral palsy. He is hypotonic when at rest and becomes wild and erratic when he attempts voluntary movement. He is also prone to having seizures which requires medication to control. His attitude is one of defiance and anger. Paul does not follow instructions well and in spite of the fact his mother was able to secure a wheelchair for him of his own, he refused to use it. When moved around at school, the clutches were disengaged and he was pushed around as if the wheelchair were manual.

Figure 12 shows daily training time over 18 training sessions. The data reveal that over the first 14 sessions Paul's training time was stable. However his performance on Sessions 13 and 14 showed a marked lack of interest and training was reduced in frequency in an effort to maintain his interest. In spite of this, his willingness to work shows a decline over the last four training sessions. Figure 13 reveals how Paul distributed his time during each training session among the three virtual worlds.

Graph of daily training time for Paul.

Figure 12: Daily Training Time for Paul.

In general, Paul's interest in working in the first virtual world peaked after six training sessions and then gradually waned until he lost almost all interest in this particular training scenario. His interest in exploring the second virtual world remains fairly stable over approximately 11 training sessions. As mentioned above, on Sessions 12 and 13 there appears to be a dramatic increase in his world 2 time but in actually Paul did nothing in terms of driving around during these two sessions. He seemed quite bored except for a few minutes spent driving in world one.

We then began to see Paul every other week in an effort to increase his motivation. The data reveal that over the last three sessions he began to work in the most difficult virtual world trying to cross the street. Probe data are revealed in Figure 14. Clearly Paul refused to perform over three consecutive probe sessions. During the last probe he demonstrated a good ability to stop before hitting a wall. This was the only time Paul was willing to work for us outside the virtual environment.

Figure 15 reveals Paul's odometer readings over the same time period. Prior to receiving virtual reality training, on 2/15/95, Paul never used his chair independently in the natural environment. Over time the data reveal Paul began using his chair appropriately and by 6/7/95 was traveling approximately 1.2 miles (1.93 km) per day. Thus it would appear that Paul's effort to drive in the virtual worlds correlates with using his chair in actual reality independently. For this young man these results are startling and dramatic.

Graph of daily training time per world for Paul.

Figure 13: Daily Training Time per World for Paul.

Graph of evaluation probes for Paul.

Figure 14: Evaluation Probes for Paul.


The data reflecting the efficacy of training motorized wheelchair operation in virtual reality are still coming in. Preliminary results are encouraging. The data indicate that as children drive in virtual reality, their performance improves in actual reality. We are encouraged in this initial effort to use this technology in this unique way. Final results will be forthcoming and will be published toward the end of 1996.

Graph of Paul's weekly mean weekly distance traveled.

Figure 15: Weekly Mean Distance Traveled for Paul.

Several concluding remarks are appropriate given our initial effort to use virtual reality technology to help children with physical disabilities and in view of other efforts to use this technology for other purposes as reported in this book of proceedings.

  • First, it is imperative that we evaluate the effects of virtual reality training whether for military applications, medical applications, or educational applications in the real world. Training an individual, whether a warrior or a child with orthopedic impairments, is of no use if the performance does not transfer from virtual reality to actual reality. To the best of this writer's knowledge, we are the first to present data which suggests generalization occurs. These are important findings.
  • Second, the world of virtual reality technology is changing very quickly. The rate of change is almost unimaginable. This makes it almost impossible for virtual reality researchers and developers to create training programs and to do research on their effectiveness before the technology upon which it is based becomes obsolete. However, it is incumbent upon us to do this.
  • Third, all of us have physical limitations in the sense that we cannot stand on the sun or shrink ourselves small enough to enter a chloroplast in the leaf of a plant. As we learn to use this technology to overcome the physical limitations of children who have severe orthopedic impairments, we will have also learned how to use this technology to overcome the physical limitations of those of us who are not disabled. In this sense, for once, the disabled will lead the way and the rest of us will follow. Our new work in science education is a step in that direction.
  • Fourth, it is easy to get lost as we are inundated with new and better technologies. An important rule to follow is this: we should never allow our tools to define our work. Our work should define the tools. We should ground our efforts in exploring this technology in the real world where needs are apparent. It is not adequate nor is it safe to start with the acquisition of sophisticated technology and then look around for an appropriate application. We, the users, will drive this field as it evolves into the future and we should remember that the cart should not go before the horse.


We gratefully acknowledge the United States Department of Education which has funded this research project for a period of three years, beginning July 1, 1993 (Grant No. H180E30001).


Acredolo, L. P. and D. Evans, 1980. Developmental changes in the effects of landmarks on infant spatial behavior. Developmental Psychology, J6: 312-318.

Adolph, K.E., E.J. Gibson, and M.A. Eppler, 1990. Perceiving affordances of slopes: The ups and downs of toddlers' locomotion. Research for the Emory Cognition Project, Emory University, Atlanta, GA.

Bertenthal, B.I., and D.L. Bai, 1987. Infants' sensitivity' to optical flow for controlling posture. Paper presented at the Biennial Meeting of the Society for Research in Child Development, Baltimore, MD.

Bertenthal, B.I., 1.1. Campos, and K.C. Barrett, 1984. Self-produced locomotion. In R.N. Emde, and R.J. Harmon (eds.), Continuities and discontinuities in development, pp.175-210. Plenum.

Held, R. and A. Hem, 1963. Movement-produced stimulation in the development of visually~guided behavior. Journal of Comparative and Physiological Psychology', 8].. 394-398.

Horobin, K., and L.P. Acredolo, 1986. The role of attentiveness, mobility history, and separation of hiding sites on Stage IV search behavior. Journal of Experimental Child Psychology, 4]: 114-127.

Kermoian, R. and J.J. Campos, 1988. Locomotor experience: A facilitator of spatial cognitive development. Child Development, 59: 908-917.


Dr. Dean Inman is a Research Scientist at Oregon Research Institute in Eugene, Oregon. In that capacity he is involved in internationally-recognized research with severely disabled children sponsored by the United States Department of Education and other Agencies. He is evaluating the possibility of using virtual reality technology to teach severely orthopedically impaired children to drive motorized wheelchairs and to develop science lessons using virtual reality to teach severely orthopedically challenged high school students basic science concepts in biology, chemistry, and physics. Dr. Inman received B.A. in general psychology with a major in physiological psychology from Sacramento State College; an M.S. in experimental psychology with an emphasis on the behavioral sciences from Utah State University; and a Ph.D. in special education and rehabilitation concentrating on skeletal muscle dysfunction in children with severe movement disorders from the University of Oregon. Over the last twenty years he has developed diagnostic and treatment techniques for individuals with severe and profound retardation and deaf-blindness and for children with severe orthopedic impairments.

Ken Loge is the implementation coordinator and designer for the Oregon Research Institute Virtual Reality Laboratories, and teaches computer media and technology courses part-time. He specializes in interactive media design and engineering, multimedia authoring, and instructional technology. Ken received degrees in broadcast media as well as in telecommunications and film, and an M.S. in instructional systems technology.

John Leavens is the lead software engineer for the Oregon Research Institute Virtual Reality Laboratories. He has a B.S. degree in applied physics from the Georgia Institute of Technology, specializing in microprocessor controls. He has worked on various projects ranging from computer vision for quality control to system libraries for computer games.