CYBERPSYCHOLOGY & BEHAVIOR Volume 5, Number 2, 2002 Mary Ann Liebert, Inc. Development and Validation of Virtual Driving Simulator for the Spinal Injury Patient JEONG H. KU, M.S., 1 DONG P. JANG, Ph.D., 1 BUM S. LEE, M.D., 2 JAE H. LEE, M.D., 2 IN Y. KIM, M.D., Ph.D., 1 and SUN I. KIM, Ph.D. 1 ABSTRACT We developed a virtual reality (VR) driving simulator in order to safely evaluate and improve the driving ability of spinal injury patients. The simulator is composed of an actual car, a beam projector, and a large screen. For the interface of our driving simulator, an actual car was adapted and then connected to a computer. We equipped the car with hand control driving devices especially adapted for spinal injury patients. A beam projector was used so that the subjects could see the virtual scene on a large screen set up in front of them. The virtual environment (VE) consisted of 18 sections (e.g., a speed-limited road, a straight road, a curved road, a left turn) and each section was linked naturally to the next. The subjects selected for this trial were 10 normal drivers with valid driving licenses and 15 patients with thoracic or lumbar cord injuries who had prior driving experience. For evaluation, five driving skills were measured, including average speed, steering stability, centerline violations, traffic signal violations, and driving time in various road conditions such as straight and curved roads. The normal subjects manipulated the gas pedal and the brake with their feet, while the patients manipulated a hand control with their hands. After they finished driving the whole course, the participants answered the questions such as How realistic did the virtual reality driving simulator seem to you? and How much was your fear reduced? In this study, we found that the difference in manipulation method (i.e., the patient group s hand control versus the normal driver s foot controls) does not seem to influence relative performance in the VR driving simulator, though training to improve the use of hand controls in the VR driving simulator would be useful to reduce the fear that the patients feel while driving. C ARS INTRODUCTION HAVE BECOME NECESSITIES of our daily life, and their importance increases from year to year. They are equally necessary for people who have difficulties in using public transportation, and they are especially important to handicapped people because they extend the range of activity and allow participation in a social life. Driving is, therefore, essential for many handicapped people in order to maintain independent lifestyles. Assessment and improvement of driving ability has become an important part of rehabilitation therapy. In many rehabilitation hospitals and centers, including the National Rehabilitation Center in Korea, movie-based driving simulators have been deployed for driver training and skill improvement because on-road tests are considered to be too dangerous to handicapped people who do not yet have sufficient 1Department of Biomedical Engineering, Hanyang University, Seoul, Korea. 2Department of Rehabilitation Medicine, National Rehabilitation Center, Seoul, Korea. 151
152 KU ET AL. skills to control a car. 1 6 Such simulators have some shortcomings in driver training because they are noninteractive. Virtual reality (VR) is a technique that constitutes a three-dimensional environment that puts the subject in a condition of active exchange with a virtual world created by a computer. A powerful advantage of the new technology is that the subject is not merely an external observer of pictures or one who passively experiences the reality created by the computer. Instead, in VR, the subject can actively modify the three-dimensional world in which he or she is acting, in a condition of complete sensorial immersion. 7 Applying a VR technique to the field of driver training can create a realistic and interactive situation for the participant. A VR driving simulator can overcome the disadvantages of on-road tests and movie-based simulators because it provides a more safe and efficient training method, which can create diverse and interactive driving situations. 8,9 The goal of this study was to implement a VR driving simulator and use it to assess and improve the driving skills of handicapped people. MATERIALS AND METHODS Driving simulator architecture A VR driving simulator is mainly composed of a hardware system and a virtual environment (VE) scenario. The hardware for our driving simulator consisted of an actual car adapted for the handicapped, interfaces between components of the car (steering wheel, brake and accelerator, directional ramp and gear) and a computer, a beam projector and screen (3 2 4 m), and a sound system (Fig. 1). We used an actual car connected to a personal computer for realism and attached a hand control device for subjects afflicted with lower limb palsy. A beam projector displayed VR situations on a screen in front of the car, and sounds were generated as appropriate for the vehicle s speed, to increase realism for the participant. The VE scenario for assessing and improving basic driving ability was based on an actual driver examination in Korea. The VE for these scenarios ran on Rhinoceros and 3D Studio Max and was developed with DirectX 7.0 and Visual C++ for real-time rendering. A city scene was created for the simulator that included S and T courses, a building, a park, tunnel road sections (amongst others), four right turns, four left turns, five traffic signals, four stop signs, two lane changes, and other scenarios. The roads in the virtual city were divided into 18 sections (Figs. 2 and 3). Subjects The subjects were 10 normal drivers (nine were male and one was female) with driving licenses and 15 patients (all male) who were in the National Rehabilitation Center in Korea for treatment of thoracic or lumbar cord injuries. FIG. 1. System architecture for virtual driving simulator.
VIRTUAL DRIVING SIMULATOR FOR THE SPINAL INJURY PATIENT 153 FIG. 2. The patient group all had prior driving experience. The ages and driving histories of the normal drivers are shown in Table 1. Among the spine injury patients, 13 had thoracic cord injuries and two had lumbar cord injuries. Also, 12 members of the patient group had driving licenses and three did not (Table 2). Road sections of driving simulator. Procedure Patients manipulated the break and accelerator pedal of the simulator by using the hand control device, whereas the normal drivers used foot controls. Subjects filled out a questionnaire that asked their age, clinical history, and period of driving absence. The subjects FIG. 3. Scene of virtual environment (in a town with a traffic signal, on a mountain road, in a tunnel, and in a city zone with a speed limit sign).
154 KU ET AL. TABLE 1. GENERAL CHARACTERISTICS OF NORMAL SUBJECTS Normal subjects (n = 10) were familiar with the road that they would drive, and they were given time (about 5 min) to become accustomed to the VE (Figs. 2 and 3). During each simulation (which took approximately 20 min), driving data were recorded by the computer. We analyzed this data and obtained evaluation parameters. For evaluation purposes, five driving skills were measured in various road conditions (such as straight road and curved road); these skills included average speed, steering stability, centerline violations, traffic signal violations, and driving time. Measurements were taken along a road that was divided into 18 sections. After subjects finished driving the whole course, they answered the following two questions: How realistic did the VR driving simulator seem to you? How much was your fear reduced in driving? RESULTS Mean (SD) Age (years) 31.4 (1.3) Driving history (years) 8.9 (3.4) In the patient group, the average period of driving abstinence after an accident was 14.5 months; and 14 members of the patient group (93.3%) manipulated a hand-control device for the first time in the experimental procedure. Measurements of the five driving skills were not significantly different between the groups (normal vs. handicapped; p > 0.05). This indicates that driving skills were not influenced significantly by the manipulating methods (i.e., hand or foot controls; Table 3). Among the 18 road sections, the average speed of the patient group was 45.6 km/h, less than the 61.2 km (p < 0.05) of the normal group in sections 1, 2, 3, 4, 8, 10, 11, 12, and 13. Thus indicates that the patients have a tendency to drive carefully. This may also be due to the fact TABLE 2. GENERAL CHARACTERISTICS OF SPINAL CORD INJURED SUBJECTS Number of cases (%) Age (years) 20 29 5 (33) 30 39 2 (13) 40 49 6 (40) 50 59 2 (13) License + 12 (80) 2 3 (20) Level Thoracic 13 (87) Lumber 2 (13) Gender Male 15 (100) Female 0 (0) that these were more challenging road sections: sections 1 and 8 were road entrances, section 2 is a speed-limited road, and sections 3 and 4 have sharp curves. The patient group also drove carefully in the T course and in the left-turn sections: 10, 11, 12, and 13. This indicates that there are no differences between the patient group and the normal group in simple road conditions, whereas there are significant differences (i.e., the patient group drove with TABLE 3. SPEEDS IN EACH SECTION (KM/H) Normal Patient p Sections (mean SD) (mean SD) value 1 36.8 (12.1) 22.6 (7.1) 0.005* 2 47.6 (13.7) 31.1 (11.5) 0.017* 3 53.4 (9.9) 33.4 (16.5) 0.021* 4 70.6 (9.6) 50.0 (20.4) 0.046* 5 57.0 (10.5) 52.5 (20.7) 0.652 6 85.0 (11.6) 62.3 (23.1) 0.052 7 58.8 (22.8) 46.8 (20.4) 0.285 8 57.6 (9.3) 52.8 (10.5) 0.213 9 52.2 (15.3) 39.9 (14.5) 0.123 10 44.2 (9.3) 26.4 (8.8) 0.001* 11 61.2 (5.6) 45.6 (13.6) 0.024* 12 30.8 (7.9) 21.8 (6.2) 0.017* 13 26.8 (7.8) 21.8 (6.7) 0.001* 14 52.2 (9.9) 51.8 6.8 0.184 15 52.6 (5.2) 52.8 15.6 0.978 16 63.0 (9.3) 48.8 16.9 0.095 17 71.2 (12.9) 58.1 19.4 0.181 18 31.8 (13.3) 22.0 4.3 0.019* *p < 0.05.
VIRTUAL DRIVING SIMULATOR FOR THE SPINAL INJURY PATIENT 155 FIG. 4. The hand control device for the handicapped. much lower speed) in road conditions that require serious driving skills. Eleven patients (73%) reported that their fear of driving was reduced when they drive with their hands; their average score on the realism question was 51.5%. This suggests that in future trials it will be necessary to enhance the degree of realism in the driving simulator. DISCUSSION The act of driving involves the interaction of cognitive abilities, sensory perception, and physical exertion with various environmental factors. Therefore, methods to investigate driving ability broadly are needed, including physical, psychiatric (i.e., cognitive and attention ability), and driving assessment (i.e., written exam, driving simulator, and behind-the-wheel road test 2 ). A driving simulator using VR techniques is very helpful in driving rehabilitation, because it can provide objective data on the participant s driving ability and skill to evaluate whether a participant should drive or not. 8 9 Although it is possible, with the VR driving simulator, to assess the ability to control a vehicle, including steering and gas and break pedal control, there are shortcomings to not investigating the interaction between the subject and other drivers. In spite of these deficiencies, there are reports that VR driving simulators are helpful in developing a minimum standard for driving fitness on-road. 10 Until now, there have been many studies to determine whether cognitive and perceptual abilities are impaired, rather than to assess the driving skills of the handicapped on the road, due to economic considerations, the time required, and road safety issues. However, these studies (during 1980 1990) were inconsistent, inconclusive, and lacked validity and practicality in their methods. 3 Interestingly, the Glaski et al. study on the assessment of driving ability focused extensively on the cybernetic model of driving in 1992. 1,5,6 They also assessed how much driving ability and FIG. 5. Driving in the simulator (left is an interior view, right is an exterior view of the car).
156 KU ET AL. technique differed according to road conditions, road type, and weather conditions using a movie-based driving simulator. 4 In the present study, the difference in manipulation methods (i.e., the patient group s hand control versus the normal driver s foot controls) does not seem to influence the relative performance in the VR driving simulator. This shows that the method of assessment with the VR driving simulator is reliable. The patients tendency to decelerate while driving in some specific conditions reflects their tendency to drive more carefully in difficult conditions. Training to improve the use of hand controls in the VR driving simulator would be useful to reduce the fear that patients feel while driving. Furthermore, 73% of the patients reported that their fear of driving dissipated after finishing their drive in the virtual city. CONCLUSION In this study, a VR driving simulator was developed and validated by clinical trial to assess and improve the driving skills of handicapped people. This demonstrated the driving abilities of patients with spinal cord injuries and helped to reduce their fear of driving with their hands. In future research, it will be necessary to enhance the realism of the simulation and increase the variety of training situations. ACKNOWLEDGMENTS This study was funded by the National Research Laboratory (NRL) Program at Korea Institute of Science & Technology Evaluation and Planning (2000-N-NL-01-C-159). REFERENCES 1. Galski, T., Ehle, H.T., & Williams, J.B. (1997). Offroad driving evaluations for persons with cerebral injury: a factor analytic study of predriver and simulator testing. American Journal of Occupational Therapy 51:352 359. 2. Katz, R.T., et al. (1990). Driving safety after brain damage: follow-up of twenty-two patients with matched controls. Archives of Physical Medicine and Rehabilitation 71:133 137. 3. Galski, T., Bruno, R.L., & Ehle, H.T. (1992). Prediction of behind the wheel driving performance in patients with cerebral brain damage: a discriminant function analysis. American Journal of Occupational Therapy 47:391 396. 4. Galski, T., Ehle, H.T., & Williams, J.B. (1998). Estimates of driving abilities and skills in different conditions. American Journal of Occupational Therapy 52:268 275. 5. Galski, T., Bruno, R.L., & Ehle, H.T. (1991). Driving after cerebral damage: a model with implications for evaluation. American Journal of Occupational Therapy 46:324 332. 6. Galski, T., Ehle, H.T., & Bruno, R.L. (1990). An assessment of measures to predict the outcome of driving evaluations in patients with cerebral damage. American Journal of Occupational Therapy 44:709 713. 7. Riva, G. (1997). Virtual reality as assessment tool in psychology. In: Riva, G., ed.virtual reality in neuropsycho-physiology. Amsterdam: IOS Press, pp. 70 79. 8. Liu, L., Miyazaki, M., & Watson, B. (1999). Norms and validity of the DriVR: a virtual reality driving assessment for persons with head injuries. CyberPsychology & Behavior 2:53 67. 9. Wald, J., et al. (2000). The use of virtual reality in the assessment of driving performance in persons with brain injury. In: Medicine meets virtual reality. Amsterdam: IOS Press, pp. 365 367. 10. Fox, G.K., Bowden, S.C., & Smith, D.S. (1998). Onroad assessment of driving competence after brain impairment: review of current practice and recommendations for a standardized examination. Archives of Physical Medicine and Rehabilitation 79:1288 1296. Address reprint requests to: Dr. Sun I. Kim Sungdong P.O. Box 55 Seoul, 133 605, Korea E-mail: sunkim@hanyang.ac.kr