Finding a target with an accessible global positioning system.
Global Positioning Systems (GPS) contain three segments: (1) the component in space, which consists of 24 operational satellites; (2) the earth-based control sites that monitor the satellites; and (3) the user-receiver component, which interprets the radio signals from the satellites (El-Rabbany, 2002). The handheld device, which is often referred to as the GPS unit, also contains Geographic Information Systems (GIS) data, which is spatial, such as the locations of roads, railroads, and other features that are commonly displayed on print maps (Taylor & Blewitt, 2006). The two main functions of GPS-GIS navigation systems are to provide users with locational information by relating the users' position, as determined by the GPS, to the map information that is stored in the onboard GIS database; and to enable users to mark specific latitude-longitude positions with electronic markers, known as waypoints, which can later be used to relocate the specific positions. Both functions are extremely useful to individuals who are visually impaired (that is, those who are blind or have low vision) and are included in the commercially available adapted GPS products designed for people with visual impairments. Map-location information could give travelers who are blind their street location whenever they requested it, and the ability to mark and relocate targets could solve the problem of locating targets that are placed in unstructured environments, that is, areas on maps that do not contain streets, such as parks, golf courses, and woodlots.
Accessible GPS devices were first introduced commercially in 2000 (Golledge, Marston, Loomis, & Klatzky, 2004). Since then, it appears that the question of whether these devices are of sufficient accuracy for use by individuals who are visually impaired has not been addressed. This research was conducted to answer that question.
On the surface, it seems that these devices are not accurate. For example, commercial GPS devices are accurate, on average, to within 9.14 meters (30 feet) (Broida, 2004). Therefore, as users walk along a sidewalk or path with a device in search of a target, they may receive information from the GPS that the target is next to them, when it is more than 9 meters away. Since a 9-meter error could place a user in the center of a street instead of at the curb ramp, one would likely question the usefulness of GPS for travelers who are blind or have low vision. However, the 9-meter error reported for GPS devices does not necessarily equate with missing a target by such distances. Most important, the devices are used in conjunction with, not in the absence of, the myriad orientation and mobility (O&M) skills that are commonly taught to people with visual impairments (Jacobson, 1993; LaGrow & Weessies, 1994; Long & Hill, 1997). If travelers in the foregoing example were using GPS navigation units that were reporting the name of and distance to the upcoming cross street as they traveled along the walk, they could use cane skills to search for the curb after the device reported the distance as 100 feet or less, and they could stop when they found the curb.
The BrailleNote GPS (BGPS), which was investigated in this research, has both the map-based orientation and the location-marking functions described earlier. However, since this study was aimed at accuracy, it focused only on marking and finding targets in unstructured environments. Two BGPS functions can be used to mark and relocate targets: the manual-route and the points-of-interest functions. The manual-route function enables users to create a series of waypoints that serve as step-by-step "breadcrumb" markers to get from one point to another along a series of waypoints. The points-of-interest function enables the user to mark any single point, such as a favorite shop, a campus building, or a favorite piece of sculpture. If an individual who is blind wants ongoing access to a large sculpture that is set in the center of an open area, he or she could mark it with a waypoint and then use the waypoint to relocate the piece to enjoy at a later time. In this example, when the sculpture waypoint is called up from the memory of the BGPS, the information given to the user as she or he travels toward the sculpture can be spoken in spatial language, that is, the clock direction and distance in feet to the target. For example, if the user was walking directly toward the sculpture, the BGPS would report, "80 feet at 12 o'clock." When the user was 20 feet closer, it would continue, "60 feet at 12 o'clock" and so on until it spoke "arrived your destination" at approximately 45 feet from the sculpture, which is likely too far away to localize it using its acoustic properties (Wiener & Lawson, 1997).
A technique that we developed, called geotracking, can be used to get extremely close to the target after the "arrived your destination" announcement is made by the BGPS when one reaches a distance of approximately 45 feet from a previously entered point of interest. It is based on the fact that if users do not stop, but continue to move toward the target after the initial "arrival" message is received, the direction and distance readings of the navigation device are more accurate than if the user is standing still. Thus, a user can be trained to continue toward the target after the arrival announcement is made, to keep striking the keys that give target distance and direction, and to contact or closely pass the target. Sometimes, the technique may require two or three "passes" before the target is located. If the target is not found on the first pass, the user continues on the 12-o'clock path for another 30-40 feet, turns 180 degrees, and makes an identical pass back toward the target. Any object that is large enough to reflect sounds, such as a vehicle, a bus stop shelter, or a mailbox, will usually be found on the first pass because geotracking takes the traveler close enough to the object to hear sounds reflecting from it or to hear the "sound shadow" that is created by the object. As far as is known, there has been no published research on the degree of accuracy of using geotracking and GPS devices. Therefore, the purpose of the research presented here was to determine the accuracy of BGPS and the geotracking technique as aids in locating objects in environments where no other cues are available for locating targets.
Two experiments were conducted to determine the participants' accuracy in locating targets using a BGPS and the geotracking technique. The first used a within-subjects group design to measure the ability of 19 participants to locate a 25-foot chalk circle from a distance of 13.6 meters (150 feet) in a paved parking lot. The second experiment, which was designed as a follow-up to the first, used a single-subject A-B-A-B research design (Gay & Airasian, 2003) to determine how close an experienced user could come to a known target point in a paved lot using the BGPS. The first experiment was designed to measure the abilities of novice users, and the follow-up experiment tested an experienced user's skills to pinpoint the possible degree of accuracy.
The participants were 19 adults, aged 20-64, who had received long cane training as university students or rehabilitation consumers. Ten participants were sighted and were or had been enrolled in an O&M-degree program at a university. Of the 9 participants with visual impairments, six were functionally blind.
Location and apparatus
Experiment 1 took place in a parking lot that was roughly rectangular (154 meters by 46 meters, or 500 feet by 150 feet). An 8.3-meter (25-foot) chalk circle was drawn near the center of the site, and an electronic waypoint was recorded at its center. Three locations at which the participants would begin to walk toward the circle were established 46 meters (150 feet) from the edge of the circle and separated from one another by 9.2 meters (30 feet). These starting points were numbered 1, 2, and 3 and were designated by 0.6-by-0.6 meter (2-by-2 foot) carpet squares, which were placed on the starting points facing the circle. The BrailleNote PDA running Keysoft 5, a braille keyboard, Version 2 BGPS software, and a Magellan GPS receiver (Model IEC-529 IPX7) were used.
Measures and procedures
After the participants read and signed the consent form provided for this project by the Western Michigan University Human Subject Institutional Review Board, their vision was occluded using taped ski goggles, and the site was described to them. They were told that their tasks would be to receive training and then to make six attempts to locate the circle by walking to it from the carpet squares without their canes and under the protection of an O&M instructor. They were also told that three of these trials would be done under the condition of using a BGPS, and three would be done using their own navigational skills. In both cases, they would be taken to the carpet square, turned to face directly toward the circle, asked to point to the circle, and then asked to begin to walk whenever they were ready.
Training assignments were made in alternating order, with even-numbered participants receiving GPS-geotracking training first and no-GPS training second, and odd-numbered participants receiving training in reverse order. The GPS-geotracking training was considered complete when we were satisfied that the participants understood the concepts of locating waypoints using clock-face directions from the BGPS and geotracking, which, as was described earlier, refers to the ability to localize a target using the BGPS after the device speaks the initial "arrived" notice. The mean number of trials required to achieve the goal of GPS-geotracking training was 2.0 for the no-GPS condition and 5.I for the GPS condition. Training for the no-GPS condition involved informing the participants of the task to be completed and asking them to practice walking to the target after being assisted in facing the target directly. Training in this condition was a simple matter of ensuring that the participants understood that they needed to walk in a line as straight as possible and to ask for assistance when they felt disoriented or thought they were well past the target. Thus, the training was considered completed when the participants finished a practice trial by locating the circle or requesting guidance or when they had no further questions.
Training for the GPS condition consisted of instructing the participants how to keep the target at the GPS position of 12 o'clock and to understand the concept of headings. Conceptualizing headings was the most difficult part of the training regimen. The difficulty stemmed from the participants' expectation that the BGPS heading function is based on a compass. Since the BGPS measures the linear path of the GPS receiver, rather than magnetic north, accurate reports of the BGPS clock-face direction are dependent on the participants taking four or five steps along a continuous linear path. The participants were taught that if they stood in one spot and rotated, the clock direction reported by the BGPS would be incorrect, remaining at the same clock position as when the rotation began, regardless of the degree of rotation. For example, if a participant was facing the target and heard "12 o'clock" from the BGPS, rotated 90 degrees clockwise, and then requested a second reading, the BGPS would say, "12 o'clock" even though the target would actually be at 3 o'clock. As a result, it took more than twice as much time to prepare the participants for the BGPS than for the no-GPS condition.
The participants were rotated between the use of the BGPS and the starting position on the carpet squares. For example, the first participant began with the BGPS on starting point 1 and proceeded to starting points 2 and 3. The second participant began with no GPS on starting point 2 and then proceeded to starting points 3 and 1. When the participants were positioned correctly and were able to point to the center of the circle, they were told which condition would be operative (GPS or no GPS), reminded that they were facing the circle and that it was 150 feet away, and asked if they had questions. Because it was possible for the participants to become significantly disoriented at 150 feet, particularly in the no-GPS condition, the participants were told that if they felt hopelessly lost, they could solicit aid from a researcher, who would give them the clock direction and distance to the target. However, such a solicitation was recorded as a failure to locate the circle independently. We stopped the participants when they stepped into the circle, walked to the outer edge of the parking-lot test area, or reached the time limit of three minutes. Data were analyzed using Wilcoxon matched-pairs statistics.
The results demonstrate a significant effect of BGPS-geotracking on locating targets (Z = -3.89, p < .01), as evidenced by the fact that 19 participants located the circle independently 53 out of 57 times (93%) while using the GPS, but only 7 out of 57 times (12%) in the no-GPS condition.
Since Experiment 1 demonstrated that the participants were highly successful in locating a 25-foot-diameter circle, the next logical question appeared to be how accurately someone could use geotracking and the BGPS. It seemed that the most logical way to determine the answer would be to test a person who was highly experienced with the technique, which led to the following experiment.
To determine the highest performance level of the BrailleNote GPS, one highly experienced GPS user was selected as the participant. He is one of the authors of this article, was the developer of the geotracking technique, is a dog guide user, is 62 years of age, has an acquired vision loss, is functionally blind, and has more than three years of experience using the BGPS.
Location and apparatus
The experiment was conducted in a roughly rectangular empty parking lot that measured 30.5 meters (99 feet) wide by 183 meters (584 feet) long. The testing area was set up much like it was in Experiment 1; that is, there were three starting points that were separated by 9.2 meters (30 feet) and were marked by carpet squares. However, the starting points differed from Experiment 1 in that they were placed at various distances (46 meters, or 150 feet; 69 meters, or 225 feet; and 92 meters, or 300 feet) from the target point. A target measuring 30.48 centimeters by 30.48 centimeters (1 foot by 1 foot) was placed in the middle of the southwest end of the parking lot. A nail was hammered into the asphalt to mark the center of the target, and the target was then outlined with white masking tape to increase its visibility to the researchers. Also, an existing white painted line extended through the target 15.2 meters (50 feet) on either side of the target and was perpendicular to the participant's line of travel. The weather was clear and a comfortable 24 degrees Celsius (77 degrees Fahrenheit), and there were no sound sources in the environment that were loud enough to serve as directional cues during the testing period. The apparatus was similar to that used in Experiment 1, except that the BrailleNote Classic running Keysoft 6.11 and BGPS version 3.0 with a Magellan receiver (model IEC-529 IPX7) was placed on the participant's shoulder.
Measures and procedures
The experiment used an A-B-A-B single-subject design, in which the A phases represented those in which the participant used only his natural skills (no GPS) and the B phases represented those in which he used the BGPS. During each phase, a minimum of three trials were used, one being initiated from each of the three starting points. The starting points were designated SP 1, SP 2, and SP 3, and the participant used them in numeric order after the first trial was selected randomly. Before the first A or baseline phase began, the participant was guided around the perimeter of the test area to establish its size. He was also given the same human-subjects protection provided in Experiment 1. After the participant was oriented to the parking lot, the first baseline phase (no BGPS) was begun by guiding the participant to the first carpet square, and a researcher standing at the target created a sound cue by clapping loudly. The participant indicated the target direction by pointing, and if his direction of pointing was incorrect, it was corrected by verbal instruction. Once he was facing the target, the clapping stopped. Preliminary tests demonstrated that the target's small size and extreme distance (up to 450 feet) from the participant made the sound cue necessary. The participant was told to try his best to walk to the point where the sound cue had originated and to proceed when ready. The dependent variable was the error distance, that is, the distance measured between the point at which the participant crossed the white painted perpendicular line to the target itself. The participant was given the cue, "stop" when he transected the white line, and the point was marked for later measurement with masking tape containing phase and trial identifiers. The A phase required four trials that used three starting points, with one point used twice. Four data points were needed in the A phase to meet the single-subject phase-change requirement of establishing negative performance in slope, where slope is the relationship between the last two data points.
After ensuring that the BGPS was working properly, the participant was taken to the selected starting point, but no sound cues were provided at the target point. The instructions for the B phase differed slightly from those for the A phase in that the participant was given two minutes to locate the target. This was an adequate amount of time to pass the target, turn, pass it again, and perhaps even pass it a third time, thus using the geotracking strategy described earlier. The dependent variable was also error distance, but in this case, it was the distance in meters measured from the nearest point to which the participant passed the target to the target itself. The participant was told to proceed when ready, the researchers marked the points at which he passed the target with identifying tape and stopped him at the end of the two-minute period. As in the baseline phase, measurements were taken after the trial was completed. The remaining A2 and B2 phases were conducted identically to A1 and B 1 phases, except that the orientation activities in the first phases were unnecessary for the second.
The results of this experiment are depicted in Figure 1. The differences in the participant's performance when using and not using the BGPS are striking, since the A phases demonstrate the participant's inaccurate and erratic performance using his own skills, but coming within 30.5 centimeters (1 foot) of the target in every B phase trial. Note the following characteristics of Figure 1:
Slope. Slope represents the relationship of the last two points within a given phase. The upward slope (poorer performance) between Trials 3 and 4 in both A phases indicates that the intervention with the BGPS was begun following a drop in performance, which meets the requirement of the experimental design.
[FIGURE 1 OMITTED]
Level. The most important analytical point is the difference in level between the data points in the A phases and the B phases. Since lower-level data points--that is fewer errors--indicate high performance, it is obvious that using the BGPS is superior for this skilled individual. All B phase trials demonstrated that the participant came within a foot of the target.
Consistency. Consistency was also markedly better in the B phases than in the A phases, as indicated by the fiat line connecting the B-phase data points and the erratic nature of the A-phase points, which ranged in error distance from 1.3 meters (4 feet) to more than 37 meters (120 feet).
Trend. No trend in the participant's performance was found from Trial 1, B1, to Trial 3, B2, since the line connecting them shows no increase or decrease in the participant's B-phase performance across the six trials. Instead, the performance was identical across all the trials. The data also demonstrate superior performance numerically. The mean error distance using the BGPS was 0.31 meter (1 foot), but 16.4 meters (53.3 feet) when the participant used only his wayfinding skills.
The data demonstrate that the novice users in Experiment 1 were capable of locating the 25-foot chalk circle with minimal BGPS-geotracking training and that the experienced user in Experiment 2 was able to pass within 1 foot of the target in every trial using the BGPS and geotracking technique. The results strongly indicate that the addition of the geotracking-training technique for locating targets decreased the 9-meter error rate reported for commercial GPS units to a level that would appear to make the device practical for locating relatively small targets. In the research, the targets were two dimensional and thus gave the participants no acoustic cues to their whereabouts, yet the experienced participant consistently came within 1 foot of them. The practical implications of this research would appear to be relatively clear: People with vision that is too limited for locating such objects as mailboxes, tent sites in campgrounds, cemetery plots, dumpsters in parking lots, works of art on large lawns, and countless other things that are not easily found using trained or natural skills can access these things through GPS-geotracking. Furthermore, it can be assumed that three-dimensional objects would be more easily located than the test targets, since they would be expected to furnish some acoustic information to users.
The research presented here had some inherent weaknesses and is surely not complete. As we noted earlier, since the participant in Experiment 2 was part of the study-design team, his familiarity with the study could have had confounding effects on the results. Despite this weakness, the team thought that the risk was acceptable, since the goal was to determine the optimal performance of the technique and device and the participant was the only person available in the geographic area with the desired level of skill. In addition, since only one of the three commercial GPS brands was tested, older versions of the BGPS software were used, and larger groups of participants would yield more data and likely lead to further instructional methods. More research is needed. The next step would seem to be to repeat Experiment 2, but using a group design and an actual object to be located in the unstructured test environment.
We hope that this research has demonstrated that the error distances reported in the literature will now be put into context. Furthermore, we hope that the importance of training that was demonstrated in the research will spur the inclusion of GPS training in O&M programs across the United States and around the world.
The authors thank student researchers Frank Murdock and William Noll, who carried out a portion of this project in partial fulfillment of their master's degrees. The work on which this article was based was supported by NIDRR Grant SB020101, awarded to Sendero Group, Western Michigan University, and other partners.
Broida, R. (2004). How to do everything with your GPS. Emeryville, CA: McGraw-Hill/ Osborne.
El-Rabbany, A. (2002). Introduction to GPS: The global positioning system. Norwood, MA: Artech House.
Gay, L. R., & Airasian, P. (2003). Educational research: Competencies for analysis and applications (3rd ed.). Columbus, OH: Merrill.
Golledge, R. G., Marston, J., Loomis, J. M., & Klatzky, R. L. (2004). Stated preferences for components of a personal guidance system for nonvisual navigation. Journal of Visual Impairment & Blindness, 98, 135-147.
Jacobson, W. H. (1993). The art and science of teaching orientation and mobility to persons with visual impairments. New York: AFB Press.
LaGrow, S., & Weessies, M. (1994). Orientation and mobility: Techniques for independence. Palmerston North, NZ: Dunmore Press.
Long, R. G., & Hill, E. W. (1997). Establishing and maintaining orientation for mobility. In B. Blasch, W. Wiener, & R. Welsh (Eds.), Foundations of orientation and mobility (2nd ed., pp. 3959). New York: AFB Press.
Taylor, G., & Blewitt, G. (2006). Intelligent positioning: GIS-GPS unification. Chichester, England: John Wiley & Sons.
Wiener, W., & Lawson, G. (1997). Audition for the traveler who is visually impaired. In B. Blasch, W. Weiner, & R. Welsh (Eds.), Foundations of orientation and mobility (2nd ed., pp. 104169). New York: AFB Press.
Paul E. Ponchillia, Ph.D., professor, Department of Blindness and Low Vision Studies, Western Michigan University, 1903 West Michigan Avenue, Mail Stop 5218, Kalamazoo, MI 49008-5218; e-mail: <firstname.lastname@example.org>. Nancy MacKenzie, M.A.; mailing address: 181 West 21st Street, Holland, MI 49423. Richard G. Long, Ph.D., associate professor, Department of Blindness and Low Vision Studies, Western Michigan University; e-mail: <email@example.com>. Pamela Denton-Smith, B.S., lead teacher, Program for the Blind and Visually Impaired, K-12; mailing address: 1458 Woodside Drive, Winston Salem, NC 27106; e-mail: <pamela.d.smith@ wmich.edu>. Thomas L. Hicks, M.A., CVRT, coordinator, Visual Impairment Team, Carl T. Hayden Veterans Administration Medical Center; mailing address: 4443 East Melrose Street, Gilbert, AZ 85297; e-mail: <thomas.hicks@va. gov>. Priscilla Miley, B.A.; mailing address: 8897 Vaughan Street, Detroit, MI 48228; e-mail: <firstname.lastname@example.org>.
|Printer friendly Cite/link Email Feedback|
|Author:||Ponchillia, Paul E.; MacKenzie, Nancy; Long, Richard G.; Denton-Smith, Pamela; Hicks, Thomas L.; Mil|
|Publication:||Journal of Visual Impairment & Blindness|
|Date:||Aug 1, 2007|
|Previous Article:||The production of brailled instructional materials in Texas public schools.|
|Next Article:||Communication with patients who have low vision.|