Volume 89, Issue 5 , Pages 996-1003, May 2008
Development of a Wheelchair Virtual Driving Environment: Trials With Subjects With Traumatic Brain Injury
Article Outline
Abstract
Spaeth DM, Mahajan H, Karmarkar A, Collins D, Cooper RA, Boninger ML. Development of a wheelchair virtual driving environment: trials with subjects with traumatic brain injury.
Objective
To develop and test a wheelchair virtual driving environment that can provide quantifiable measures of driving ability, offer driver training, and measure the performance of alternative controls.
Design
A virtual driving environment was developed. The wheelchair icon is displayed in a 2-dimensional, bird's eye view and has realistic steering and inertial properties. Eight subjects were recruited to test the virtual driving environment. They were clinically evaluated for range of motion, muscle strength, and visual field function. Driving capacity was assessed by a brief trial with an actual wheelchair. During virtual trials, subjects were seated in a stationary wheelchair; a standard motion sensing joystick (MSJ) was compared with an experimental isometric joystick by using a repeated-measures design.
Setting
Subjects made 2 laboratory visits. The first visit included clinical evaluation, tuning the isometric joystick, familiarization with virtual driving environment, and 4 driving tasks. The second visit included 40 trials with each joystick.
Participants
Subjects (n=8; 7 men, 1 woman) with a mean age of 22.65±2y and traumatic brain injury, both ambulatory and nonambulatory, were recruited.
Interventions
The MSJ used factory settings. A tuning program customized the isometric joystick transfer functions during visit 1. During the second visit, subjects performed 40 trials with each joystick.
Main Outcome Measure
The root mean square error (RMSE) was defined as the average deviation from track centerline (in meters) and speed (in m/s).
Results
Data analysis from the first 8 subjects showed no statistically significant differences between joysticks. RMSE averaged .12 to .21m; speed averaged .75m/s. For all tasks and joysticks, driving in reverse resulted in a higher RMSE and more virtual collisions than forward driving. RMSE rates were greater in left and right turns than straight and docking tasks.
Conclusions
Testing with instrumented real wheelchairs can validate the virtual driving environment and assess whether virtual driving skills transfer to actual driving.
Key Words: Brain injuries, Rehabilitation, Wheelchairs
PERSONAL-POWERED MOBILITY is recognized as a crucial component of rehabilitation after severe physical disability. The ability for a nonambulatory adult to drive an electric-powered wheelchair (EPW) functionally and unsupervised in a variety of environments is a significant determinant for employability, community access, and self-esteem.1 Unfortunately, not all persons who desire powered mobility can show the ability to drive safely during clinical evaluation. Fehr et al2 reported in a survey study that between 20% and 40% of individuals seeking powered mobility could not be served because of diminished upper-limb motor control, sensory limitations, and cognitive impairments. Fehr concluded that improved control systems could benefit half of these subjects. A second reason clinicians hesitate to recommend an EPW is the limited time available during the clinical evaluation to determine driving skills and the cost and risks of providing real-world training. Virtual reality (VR) has attracted much interest as an evaluation and treatment tool. VR offers protection from real-world hazards and offers objective outcome measures. Several wheelchair simulators have been prototyped, and informal descriptions can be found in conference proceedings.3 At least 1 U.S. patent has been issued for a wheelchair-training platform.4 Little hypothesis-driven, peer-reviewed research has been published on the feasibility of using VR to train people on how to drive EPWs. Victorien et al5 searched multiple medical databases back to 1975 and located only 5 journal articles on this subject, and only 2 of these addressed populations with neurologic injuries.6, 7
The study by Webster et al6 reported successful outcomes with 40 stroke patients, but the protocol and instruments were not well defined. The real-world portion of this study was apparently performed with manual wheelchairs, but the computer-simulation activities were performed with a 4-button direction selector. The study by Harrison et al7 did use EPWs and joysticks as inputs to the VR program. However, only a cursory description of the VR system and the data collection methods are provided. Results were presented as 6 single-subject case studies with mixed results.
Research Objectives
As a center of excellence for wheelchairs and rehabilitation engineering, one of our laboratory's key research activities is making powered mobility accessible to populations who are underserved.
The Human Engineering Research Laboratories (HERL) has conducted prior studies investigating whether isometric joysticks might provide a better control interface for people with spinal cord injury8, 9, 10 and neuromuscular disorders.11, 12 The instrumentation for our earlier studies involved considerable preparation and postprocessing. Tracking live wheelchairs required placing fiducial graphics on the driving surface and recording them with a video camera mounted on the back of the moving chair. Although better motion-tracking technology is now available, it is still too complex for clinical or community use. Advances in computer hardware, however, have made wheelchair virtual driving environments practical. Virtual driving environment instruments can provide real-time measures of wheelchair driving skills and provide a means of evaluating alternative joystick controls and transfer functions.
Research Population
Our first research use of our current virtual driving environment was with subjects with traumatic brain injury (TBI). TBI is a major cause of morbidity and mortality throughout the world. In the United States alone, 1.1 million emergency department visits, 235,000 hospitalizations, and 50,000 deaths occur annually.13 The prevalence of U.S. TBI survivors with long-term disability is 5.3 million14 but is based on 1996 data. Military casualties have also increased startlingly in the past few years; 6 of every 10 injured U.S. soldiers returning from recent military conflicts are diagnosed with TBI.15, 16 Based on our recruitment sample, the number of TBI survivors who use EPWs is small. Only one of the 8 subjects relied on an EPW. However, the frequent outcomes of TBI including retarded motor function, inattention, and visual field loss may easily eliminate a TBI survivor from consideration for a power wheelchair. This study reports on data collected from the first 8 subjects who have completed the phase I virtual driving trials.
Methods
Instrumentation
To collect objective data and evaluate alternative controls, we developed a virtual driving environment with interfaces for a conventional motion-sensing joystick (MSJ) and our experimental isometric joystick. We also embedded our customized algorithms in the virtual driving environment software and wrote a tuning program to aid in selecting the algorithm parameters.
Virtual Driving Environment
Figure 1 shows the virtual driving environment developed at the HERL. Driving tasks are rear projected on a 1.8-m (6-ft) tall and 2.4-m (8-ft) wide screen. The subject is seated in a stationary wheelchair. An isometric joystick (shown) or a traditional MSJ can be mounted in the quick-release cradle. The investigator manages the virtual driving environment from a laptop computer. The virtual driving software was developed by using DirectX.a DirectX provides a 640×480 pixel screen resolution that we scaled to represent a real-world area of 16.2×12.2m. The wheelchair icon represents a 71×89cm rear-wheel drive wheelchair. All results in this study are stated in equivalent real-world dimensions to maintain consistency. To model EPW steering, acceleration, and braking, we mounted 2 high-resolution data loggers on a Quickie P300 EPW17 and recorded ballistic samples. A quasi-proportional derivative model was used in the simulator's update functions. The resulting wheelchair icon moves realistically in both forward and reverse directions. Moving the joystick 90° left or right of center (9 o'clock or 3 o'clock) causes the wheelchair icon to rotate about its center. When driving in reverse and turning (8-o'clock or 4-o'clock joystick positions), the chair moves in the correct arc, toward the operator's left when applying 4 o'clock and to the right when applying 8 o'clock. Note that this is the opposite of automotive steering.
Fig 1.
The virtual driving environment. Subjects sit in a wheelchair; the head position monitor surrounds the back of the head and the isometric joystick or MSJ is adjusted for comfortable access. The display is a 1.2×2.4m (4×8ft) rear-projection screen. The racetrack task is used for familiarization. The investigator's laptop computer is visible in the foreground.
Joysticks Controls
An MSJ is a fairly intuitive control; the handle pivots outward from a spring-loaded center, the chair moves approximately in the direction the handle is pointed, and velocity is proportional to the magnitude of the stick inclination (typically 0°–18°). People with TBI may lack the motor skills to functionally operate an EPW because of tremors, spasticity, weakness, and attention deficits.18 An alternative to the MSJ is an isometric joystick; instead of motion, the magnitude and direction of force applied by a user's hand serves as input. Our laboratory has been researching isometric joysticks since the late 1980s. In our current design,19 the handle of the isometric joystick is mounted on a rigid stainless steel post. Forces are measured by 2 separate strain-gauge bridges bonded to the lower end of the post. A microcontroller inside the joystick enclosure samples the strain gauges, performs analog to digital conversion, applies transfer functions, and lastly performs a digital to analog conversion so that the output signals match the format of a commercial movement joystick. Providing hardware transparency allows side-by-side trials. Either joystick may be connected to drive the virtual driving environment or an EPW.
Software to Enhance Wheelchair Control
Control systems on commercial EPWs can be programmed to modify their responsiveness to user input and set maximum driving velocities and turn rates. These programming features retard EPW performance to a level the clinician feels the consumer can safely manage. Adjustable algorithms or adaptive filters that might improve a consumer's motor performance are not available. Algorithm enhancement of an MSJ is limited because the feel and responsiveness are determined by the centering spring and mechanical boundaries that guide the stick motion. Features such as dead zone, gain, bounding template, and axes rotation are mechanically defined and require mechanical modification to customize. Because an isometric joystick post is immobile, our laboratory has developed a set of compensation algorithms that emulate these mechanical features in software and allow them to be individually customized.20 These algorithms were installed on our virtual driving environment platform. We developed a tuning program to customize the algorithm parameters for each subject. A description of the tuning software procedures appears in the Protocol section.
Head Position Monitor
The head position monitor in figure 1 was developed for phase II of this study in which subjects pilot actual chairs. The head position monitor is a safety feature; if the orientation of the driver's head is not synchronized with the direction of travel selected on the joystick, it is presumed that the subject is not looking where he/she is going and either a virtual or real wheelchair can be halted.21 The head position monitor tracks head position by using a fixed array of analog Hall effect sensors triggered by a small magnet embedded in a head band. The head position monitor had only a minor role during phase I because the subject's head is in a neutral position most of the time when performing virtual driving on a flat screen; analysis of the head position monitor data will be deferred to a separate article.
Participants
Subjects were adults with TBI, 18 to 80 years of age, and at least 1 year postinjury. Subjects with and without ambulatory impairments were recruited but had to have sufficient hand function to operate an armrest-mounted joystick. We are reporting on the 8 subjects who have fully completed the phase I protocol. Seven of the subjects were men. One used an EPW, 3 used manual wheelchairs or walkers, and the remaining 4 were ambulatory. The mean age was 22.65±2 years; almost all had incurred their TBI from motor vehicle collisions when they were teenagers. All were able to comprehend and discuss the risks and benefits of the protocol meeting the requirements of our institutional review board (IRB) for informed consent.
Protocol
On arrival at the study facility, participants were greeted and the testing activities described. Inclusion and/or exclusion requirements and the capacity to consent were reviewed; subjects signed an IRB-approved informed consent. Subjects were interviewed about their general health and circumstances of their TBI. Subjects underwent a mat assessment to document range of motion, sitting dimensions, and postural support needs. Subjects were observed driving an actual EPW, a personal chair, or a loaner; the subject had to be able to drive the wheelchair indoors for 5 minutes, execute left and right turns on request, make 3 point turns, and back through a 91-cm wide doorway. Each subject had to complete these maneuvers without striking objects or requiring more than 1 or 2 verbal cues. All 8 subjects met these criteria.
Tuning Software Procedures
Figure 2 is a screenshot of the tuning program used to select the isometric joystick algorithm parameters. The algorithms were customized for each subject. First, subjects rested their hands on the joystick handle and were instructed not to apply any force. The tuning software collected x and y signals for 30 seconds and scatter plotted the xy pairs on screen to indicate a zone of involuntary movement. The investigator used a mouse to draw a circle, ellipse, or rectangle around the scatterplot to define a dead zone; signals generated within the dead-zone region are blocked during driving trials. Next, each subject was instructed to push the stick in each of the 4 principle directions for 2-minute intervals. They were instructed to select a force they could comfortably maintain, one they would prefer to use on an actual EPW to cruise along an unobstructed, level surface. A bar graph provided feedback; it turned red if the subject applied very low or very high forces outside the isometric joystick's response range; otherwise, the bar remained green. The four 2-minute datasets were plotted over the dead zone. The investigator drew a second geometric shape around the 2-minute samples to define a template, the software equivalent of an MSJ stop ring. Force vectors entered by the operator that exceeded the template were automatically scaled back to the template perimeter by the template algorithm during driving trials. The tuning program automatically calculates x and y axis gains needed so the subject's self-selected cruising speed will drive the wheelchair icon to a maximum scaled speed of 1.83m/s (≈4mph).
Fig 2.
A screenshot of the tuning software used to configure the isometric joystick algorithms. A 30-second sample of the hand at rest creates the central bit plot. Four 2-minute samples at cruising speed on the 4 principle directions generate the cross bit plot. The circle and ellipse represent the best-fit transfer functions consistent with the data.
The left portion of figure 2 shows a plot from a subject with adequate hand control. To continue in this study, a subject's level of control had to meet the following criteria: (1) the dead-zone major axis must be less than 40% of the user's maximum force; (2) the bias axes, although not required to be 90° apart, had to have at least 45° of separation to discern whether the person intended to drive left, right, or straight; (3) the mean force recorded during the comfortable force test must be at least 50% larger than the dead-zone force; this statistic was required to discern whether the person intended to drive or not; and (4) the maximum force exerted in any direction must be less than 50% of the yield strength of the stick to minimize the risk of a mechanical failure. The tuning software automatically calculated these criteria and indicated whether the subject showed sufficient hand control. Subjects were provided up to 3 tuning attempts before being terminated from the study.
Visit 1 Trials
The 4 driving tasks are shown in figure 3. The track widths of the left and right turn tasks are 3m with dashed guidelines set 1m apart. During the straight task, the track narrows from 3 to 2m. The docking task begins and ends with a 2-m track that narrows to 1.6m during an S-curve maneuver. If the wheelchair icon crosses a track edge, a virtual boundary violation occurs. After a violation, the program beeps, the wheelchair icon changes color briefly, a boundary violation counter is incremented, the software moves the wheelchair icon back to the center of the track, and the trial continues.
Fig 3.
The 4 driving tasks used in the TBI trials: left turn, right turn, straight ahead, and docking. The tasks are presented randomly and performed in both forward and reverse directions.
After algorithm fitting, subjects completed 8 driving task trials by using the isometric joystick. Each driving task was performed once in the forward and reverse direction. The subjects were instructed to drive as accurately as practical and to maintain the virtual chair within the dashed markings on all tasks. These tasks completed the first visit.
Visit 2
The isometric joystick algorithms settings determined during visit 1 were reinstalled, and the subjects were allowed to make several practice laps on the racetrack screen (see fig 1); adjustments were made to the algorithms if needed. Either the MSJ or isometric joystick was randomly selected and mounted on the quick-release cradle. The subject completed 2 sessions of 40 trials, 4 driving tasks by 2 directions by 5 repetitions. Balanced randomization was used to sequence the trials. Subjects were offered a rest break while we installed the second joystick; subjects self-selected the amount of rest they wanted, typically 2 to 10 minutes. The 40-trial protocol was repeated by using the other joystick. These 80 trials comprised the second visit.
Data Collection
A data file was created for each trial containing all settings, summary measures, and time series data. Settings include the subject's identifier code, the joystick type, algorithm parameters, the driving task, and driving direction. Summary measures included completion time in milliseconds and number of boundary contacts during the trial. The time series database contained fields for the joystick's speed and direction input signals and the wheelchair icon's x and y coordinates, the rotational orientation, speed, and acceleration. The display image was refreshed every 17ms (≈58Hz). During each refresh, a new record was appended to the database.
Data Analysis
The raw data files were processed to compute the 2 variables of interest: root mean square error (RMSE) and average speed. RMSE is defined as the average absolute orthogonal deviation of chair from the center of track and calculated by using the following equation.
(1)While calculating RMSE values, the raw position data were filtered so that only the first occurrence of repeating coordinate pairs was included. This was done to prevent biasing the RMSE with repetitious values collected while the wheelchair icon was stationary. Average speed was computed by taking the mean of all speed values generated during the trial.
The virtual driving performance of participants was operationally defined as a combination of both RMSE and average speed variables. RMSE and average speed values from the 5 repeated trials were averaged and used for analysis. Data were divided for forward and reverse driving directions and were analyzed separately. A 2×4 (joystick by tasks) multivariate analysis of variance (ANOVA) was performed to compare the driving performance of participants between joysticks (isometric joystick vs conventional or MSJ). The 4 tasks (left turn, right turn, straight, docking) served as within-subject comparison factors. Pairwise comparisons were performed to find out specific differences between the variables of interest.
Results
Graphic Examples of Trials
Figure 4 offers examples of wheelchair icon paths for the 4 driving tasks showing the diversity of our first 8 subjects; this was most evident when first introduced to the virtual driving environment. Table 1, Table 2 list the statistical results.
Fig 4.
Example of wheelchair icon pathways and range of performance; the upper row shows good joystick control, and the lower set shows impaired joystick control.
Table 1. Differences for RMSE and Speed in Forward Direction
| RMSE (m) | Average Speed (m/s) | |||
|---|---|---|---|---|
| Tasks | IJ | MSJ | IJ | MSJ |
| 1. Left turn | 0.161±0.048 | 0.195±0.089 | 0.736±0.208 | 0.735±0.219 |
| 2. Right turn | 0.175±0.074 | 0.218±0.111 | 0.747±0.237 | 0.756±0.240 |
| 3. Straight | 0.137±0.08 | 0.135±0.064⁎† | 0.759±0.263 | 0.809±0.358 |
| 4. Docking | 0.135±0.067 | 0.121±0.051⁎† | 0.642±0.226⁎†‡ | 0.695±0.271‡ |
⁎Indicates a statistical significant difference from task 1 (left turn). |
†Indicates a statistical significant difference from task 2 (right turn). |
‡Indicates a statistical significant difference from task 3 (straight). |
Table 2. Differences for RMSE and Speed in Reverse Direction
| RMSE (m) | Average Speed (m/s) | |||
|---|---|---|---|---|
| Tasks | IJ | MSJ | IJ | MSJ |
| 1. Left turn | 0.232±0.109 | 0.221±0.069 | 0.652±0.172 | 0.739±0.167 |
| 2. Right turn | 0.206±0.089 | 0.239±0.115 | 0.692±0.205 | 0.799±0.201 |
| 3. Straight | 0.141±0.049⁎ | 0.133±0.054⁎† | 0.625±0.215 | 0.652±0.208 |
| 4. Docking | 0.159±0.069 | 0.143±0.055† | 0.558±0.171⁎ | 0.553±0.174⁎†‡ |
⁎Indicates a statistical significant difference from task 1 (left turn). |
†Indicates a statistical significant difference from task 2 (right turn). |
‡Indicates a statistical significant difference from task 3 (straight). |
Forward Direction
For the forward direction, the driving performance among joysticks was not significantly different between isometric and movement joystick (main effect) (P=.89, partial η2=.018). However, a significant difference was found in the driving performance of participants within 4 tasks (main effect) averaged across the types of joystick (P=.012, partial η2=.787). Also, no significant difference was evident in driving performance among tasks averaged across joysticks (interaction effect) (P=.943, partial η2=.147).
Univariate ANOVAs were performed to find the pattern of differences in driving performance for both the RMSE and for the average speed of travel. A statistically significant difference in the RMSE was found within 4 tasks (left turn, right turn, straight, docking) (P=.006, partial η2=.317). However, there was no significant difference for speed between tasks (P=.062, partial η2=.203).
Pairwise comparisons for the RMSE indicated that participants using isometric joysticks had an equal number of errors among all tasks. With the movement joystick, participants had the highest number of errors while making right turns (.218±.111) and the lowest for docking (.121±.051, P=.013). Also, a significant difference was determined between left turn-straight driving (P=.033), between left turn-docking (P=.022), and between right turn-straight (P=.006). This suggests that lower error values occurred while performing driving tasks with narrower pathways (straight and docking) versus wider pathways (left turn and right turn).
Reverse Direction
For the backward direction, the driving performance among joysticks did not significantly differ between isometric joysticks and MSJs (main effect) (P=.811, partial η2=.032). However, a significant difference occurred in the driving performance of participants within 4 tasks (main effect) averaged across types of joystick (P<.001, partial η2=.959). Also, no significant difference occurred in the driving performance among tasks averaged across joysticks (interaction effect) (P=.084, partial η2=.646). Univariate ANOVA indicated that the RMSE differed significantly within 4 tasks (left turn, right turn, straight, docking) (P=.002, partial η2=.376). Also, a significant difference was found for the average speed between tasks (P=.010, partial η2=.336).
Pairwise comparisons for RMSE values showed that participants using isometric joysticks had the highest number of errors while making left turns (.232±.109), whereas the lowest errors were recorded for going straight reverse (.159±.069, P=.008). For MSJ, participants had the highest number of errors during right turns (.692±.205) and the lowest for straight driving task (.558±.171, P=.009). Furthermore, the number of errors were significantly higher during left turns as compared with straight driving (P=.009) and between right turn-docking (P=.034).
For isometric joysticks, speed while docking was significantly slower than the speed during left turns (P<.001). For movement joystick, speed was highest while making right turns (.799±.291) and lowest during docking (.553±.174, P=.007). Also, there was a significant difference between left turns-docking (P<.001) and straight driving-docking (P=.008).
Discussion
The results suggested (fig 5) that participants showed a higher RMSE when driving on wider tasks (left turn and right turn) than on narrower tasks (straight and docking). This difference was significant when using an isometric joystick for reverse driving and MSJ for both forward and reverse driving. When using MSJ, the participants showed a tendency to oversteer the virtual wheelchair by applying full left or full right throttle to the joystick while making small turns or corrections to their trajectories. These jerky joystick inputs, instead of incremental inputs, deviated the wheelchair farther away from the center of the track; hence, they needed extra time to correct their trajectory. Because the isometric joystick post is rigid it has a smaller response delay compared to MSJ. This makes it easier to apply an incremental force to the joystick and hence makes isometric joysticks less sensitive to oversteering. The law of Steering22 is a variation of Fitt's law for a 2-dimensional tunnel path. The law of steering predicts the relationship between trial completion time, travel distance, and width of the tunnel relative to the width of the vehicle. According to this law, the greater the clearance between the vehicle and the track, the less time required to complete a task. This is reflected in our results from higher average speed values on wider tasks (fig 6) as compared with the narrower ones when the participants used MSJ. Future analyses on the data will check the validity of the steering law in this interface.
Fig 5.
Average RMSE of the wheelchair icon for all 4 tasks in forward and reverse orientations. Abbreviation: IJ, isometric joystick.
Fig 6.
Average speed of wheelchair icon for all 4 tasks in forward and reverse orientations.
The tasks designed for this study are representative of maneuvers a wheelchair user would typically encounter in the real world. Of the 4 tasks tested, docking appears to be the most challenging because of the narrow S turn that occurs midway through the task (see fig 3; last task on right); subjects had to maneuver carefully to avoid the boundaries. Results show that several minor statistically insignificant differences were found between the isometric joystick and MSJ. Participants using isometric joysticks showed less RMSE (driving at similar speeds) while driving in forward direction. Other differences will become more apparent on future analysis involving more participants.
In a 2-dimensional bird's eye view simulation, the joystick must be mentally rotated with the wheelchair icon. For example, when the wheelchair icon is driven horizontally across the screen, the joystick is still pressed forward not sideways. Subjects also managed to perform reverse driving even though an EPW backs to the left when the joystick is pushed to the right, the opposite of vehicles equipped with steering wheels.10 Other investigators23, 24, 25 have reported on the sparing of procedural memory after TBI; our clinical observations during this study agree with their findings.
Study Limitations
Small sample size and subjects with different levels of impairment limit us from drawing many definitive conclusions from the current analysis. Unclear is how the cognitive load of maneuvering a virtual wheelchair from a bird's eye viewpoint of the tasks would relate to the cognitive load while driving an actual wheelchair in the real world. Although the participants selected which hand to use with the joystick, the statistical analyses did not control for their handedness. We also did not control for any preexisting vision problems the participants had, although we did screen for enough visual field to complete the research. We speculate that subjects took advantage of the extra clearance space provided in the turn tasks and drove less precisely. TBI survivors operated both the familiar MSJ and the novel isometric joystick equally well in the virtual environment. By the end of the second visit, all but 1 subject were able to synchronize the joystick with the desired direction of the wheelchair icon (unlike a mouse cursor, one continues to push a joystick forward even when the wheelchair icon is moving sideways). The near universal mastering of this nonintuitive control strategy is an interesting sidebar to this study; our findings supply new evidence that motor learning remains intact after TBI.
Future Directions
The second phase of this study will recruit phase I subjects and discern whether exposure to virtual training is correlated with a real-world wheelchair driving performance and, if so, if it will enhance this real-world wheelchair driving performance. There remains the question of how well subjects with TBI can drive actual wheelchairs by using either a standard or algorithm enhanced joysticks.
Conclusions
It is noteworthy that the 8 subjects were able to functionally drive the wheelchair icon in the virtual driving environment and that their scores using the entirely novel isometric joystick control were not statistically distinguishable from the more familiar MSJ; even subjects who were totally ambulatory and had no prior experience driving wheelchairs acquired this skill. Subjects were also able to master the reference rotation required as the wheelchair icon turned corners.
Supplier
References
- . Disability statistics report: mobility device use in the United States. Washington (DC): US Department of Education, National Institute on Disability and Rehabilitation Research; 2000;
- . Adequacy of power wheelchair control interfaces for persons with severe disabilities: a clinical survey. J Rehabil Res Dev. 2000;37:353–360
- . Simulation of the behaviour of a powered wheelchair using virtual reality. In: Proceedings of the 3rd International Conference on Disability, Virtual Reality and Associated Technologies. Sep 23-25; Alghero (Italy). 2000;p. 9–14
- Wu HS, Chen JJ, Chen WL, inventors; Taiwan Bicycle Industry R&D Center, assignee. Wheelchair training platform. US Patent 7004885. 2006 Feb 28.
- . Virtual reality for mobility devices: training applications and clinical results: a review. Int J Rehabil Res. 2007;30:91–96
- . Computer-assisted training for improving wheelchair mobility in unilateral neglect patients. Arch Phys Med Rehabil. 2001;6:769–775
- . The role of virtual reality technology in the assessment and training of inexperienced powered wheelchair users. Disabil Rehabil. 2002;24:599–606
- . Force sensing control for electric powered wheelchairs. IEEE Trans Control Sys Techno. 2000;8:112–117
- . Analysis of position and isometric joysticks for powered wheelchair driving. IEEE Trans Biomed Eng. 2000;47:902–910
- . Comparison of virtual and real electric powered wheelchair driving using a position sensing joystick and an isometric joystick. Med Eng Phys. 2002;24:703–708
- . Advancements in power wheelchair joystick technology: effects of isometric joysticks and signal conditioning on driving performance. [published erratum in: Am J Phys Med Rehabil 2006;85:861] Am J Phys Med Rehabil. 2006;85:631–639
- . Force control strategies while driving electric powered wheelchairs with isometric and movement-sensing joysticks. IEEE Trans Neural Syst Rehabil Eng. 2007;15:144–150
- . Rates of hospitalization related to traumatic brain injury—nine states 2003. MMWR Morb Mortal Wkly Rep. 2007;56:167–170
- . Traumatic brain injury in the United States: a public health report. J Head Trauma Rehabil. 1999;14:602–615
- . A tale of two cities. Facets Magazine. 2006;26–27Fall/Winter
- . Traumatic brain injury in the war zone. N Engl J Med. 2005;352:2043–2047
- . Using miniature data loggers to characterize electric powered wheelchair performance. In: Proceedings of the 29th Annual RESNA Conference. Jun 22-26; Atlanta (GA). 2006;
- . Meeting the seating and mobility needs of the client with traumatic brain injury. J Head Trauma Rehabil. 1992;7:81–93
- . Evaluation of an isometric joystick with control enhancing algorithms for improved driving of electric powered wheelchairs. [dissertation] Pittsburgh: Univ Pittsburgh; 2002;
- Cooper R, Spaeth D, Guo S. Variable compliance joystick with compensation algorithms. US patent 20050195166 pending.
- . Determining head position to assist electric-powered wheelchair operation for persons with traumatic brain injury. In: Proceedings of the 28th Annual RESNA Conference. Jun 25-27; Atlanta (GA). 2005;
- . Human movement performance in relation to path constraint—the law of steering in locomotion. In: Proceedings of the IEEE Virtual Reality Conference. Mar 22-26; Los Angeles (CA). 2003;
- . Procedural memory during posttraumatic amnesia in survivors of severe closed head injury. Arch Neurol. 1989;46:911–916
- . Preserved learning and retention of pattern-analyzing skill in amnesia: dissociation of knowing how and knowing that. Science. 1980;210:207–210
- . Extending the limits of complex learning in organic amnesia: computer training in a vocational domain. Neuropsychologia. 1989;27:107–120
- a Microsoft Corp, One Microsoft Way, Redmond, WA 98052-7329.
Supported by the National Institute on Disability and Rehabilitation Research, U.S. Department of Education (grant no. H133A020502) and supported with resources and facilities by the Human Engineering Research Laboratories, VA Pittsburgh Healthcare System.No commercial party having a direct financial interest in the results of the research supporting this article has or will confer a benefit upon the authors or upon any organization with which the authors are associated.
PII: S0003-9993(08)00084-1
doi:10.1016/j.apmr.2007.11.030
© 2008 American Congress of Rehabilitation Medicine and the American Academy of Physical Medicine and Rehabilitation. Published by Elsevier Inc. All rights reserved.
Volume 89, Issue 5 , Pages 996-1003, May 2008
