Disability and Rehabilitation: Assistive Technology

ISSN: 1748-3107 (Print) 1748-3115 (Online) Journal homepage: http://www.tandfonline.com/loi/iidt20

Understanding route choices for wheelchair navigation Piyawan Kasemsuppakorn, Hassan A. Karimi, Dan Ding & Manoela A. Ojeda To cite this article: Piyawan Kasemsuppakorn, Hassan A. Karimi, Dan Ding & Manoela A. Ojeda (2015) Understanding route choices for wheelchair navigation, Disability and Rehabilitation: Assistive Technology, 10:3, 198-210 To link to this article: http://dx.doi.org/10.3109/17483107.2014.898160

Published online: 20 Mar 2014.

Submit your article to this journal

Article views: 66

View related articles

View Crossmark data

Citing articles: 1 View citing articles

Full Terms & Conditions of access and use can be found at http://www.tandfonline.com/action/journalInformation?journalCode=iidt20 Download by: [Australian National University]

Date: 05 November 2015, At: 15:51

http://informahealthcare.com/idt ISSN 1748-3107 print/ISSN 1748-3115 online Disabil Rehabil Assist Technol, 2015; 10(3): 198–210 ! 2014 Informa UK Ltd. DOI: 10.3109/17483107.2014.898160

ORIGINAL RESEARCH

Understanding route choices for wheelchair navigation Piyawan Kasemsuppakorn1, Hassan A. Karimi2, Dan Ding3, and Manoela A. Ojeda3 School of Science and Technology, University of the Thai Chamber of Commerce, Bangkok, Thailand, 2Geoinformatics Laboratory, School of Information Sciences, and 3Department of Rehabilitation Science and Technology, University of Pittsburgh, Pittsburgh, PA, USA

Downloaded by [Australian National University] at 15:51 05 November 2015

1

Abstract

Keywords

Purpose: To validate a personalized routing technique with wheelchair users, understand their route choices and acquire their feedback on the necessity of wheelchair navigation and the importance of personalized routes. Method: A routing technique using a weighting method, called Absolute Restriction Method (ARM), was employed to compute personalized routes based on users’ routing preferences. The evaluation involves five manual wheelchair users. The study protocol consists of three sessions: pre-activity, activity and post-activity sessions. The evaluation included a comparison between personalized routes and shortest feasible routes, in terms of route characteristics and users’ ratings of important parameters. Results: Subjects travelled a 14.64% longer distance along the personalized routes than the shortest feasible routes. However, all personalized routes had better path quality (slope and surface condition) than the shortest feasible routes. Four out of five subjects rated the parameters they deemed most important higher for the personalized route than for the shortest feasible route. Conclusions: The study confirmed that the shortest route criterion is not always suitable for individuals with mobility impairments. Personalized routes that take into account individual characteristics, route preferences and environmental characteristics are a promising solution to lessen the difficulties that manual wheelchair users face when navigating unfamiliar environments.

Assistive technology, global positioning system, route choices, wheelchair navigation system History Received 15 October 2013 Revised 14 February 2014 Accepted 22 February 2014 Published online 20 March 2014

ä Implications for Rehabilitation  

Wheelchair users indicate the importance of personalized routes for individuals with mobility impairments. In regard to evaluation results, although subjects travelled 14.64% more distances in average along the personalized routes than the shortest feasible routes, they rated the personalized routes better path quality and less effort to travel.

Introduction Independent mobility is one of the most important factors in quality of life for individuals with mobility impairments such as wheelchair users [1,2]. In everyday travel, wheelchair users are confronted with an array of barriers including personal, interpersonal and environmental, especially in unfamiliar environments [2–5]. The reported environmental barriers include narrow sidewalks, poor travel surfaces, lack of ramps or very steep ramps, lack of curb cuts or blocked curbed cuts, sidewalk obstructions, and bad weather [1,3]. To address the issue, the United States Access Board published new design guidelines that cover access for people with disabilities under the Americans with Disabilities Act (ADA) and cover the specific accessibility guidelines for buildings and facilities under the Architectural

Address for correspondence: Dr Piyawan Kasemsuppakorn, School of Science and Technology, University of the Thai Chamber of Commerce, Bangkok, Thailand. E-mail: [email protected]

Barriers Act (ABA) [6]. However, outdoor environmental barriers still exist and prevent wheelchair users to participate in social and recreational activities. The difficulty and danger associated with traveling might be significantly reduced if wheelchair users are provided with routes and directions that can guide them to their destinations safely, accurately and efficiently. With the success of in-car navigation systems/services [7] and the widespread use of online routing services [8–10], interest in pedestrian navigation systems/services [11] is on the rise. Similar to car drivers, pedestrians also require systems/services that can assist them in navigating urban and unfamiliar environments. While conceptually wheelchair navigation features are similar to that of car and pedestrian navigation, they are implemented differently due to distinct requirements and environmental barriers. For instance, routing for car navigation pays little attention to whether a vehicle can travel on a street with a steep slope or poor surface conditions. Even pedestrian navigation systems/services that have more in common with wheelchair navigation systems/services is less concerned about certain outdoor obstacles and accessibility issues. For safe travel,

Route choices for wheelchair navigation

Downloaded by [Australian National University] at 15:51 05 November 2015

DOI: 10.3109/17483107.2014.898160

routing for wheelchair navigation must avoid travelling on pedestrian paths with steep slopes, poor surface conditions and narrow paths. However, current navigation systems/services and existing online routing services (e.g. Google Maps, Bing Maps, MapQuest) do not provide effective guidance for wheelchair navigation since they do not account for the unique requirements and preferences of wheelchair users. As a result, several research projects addressing the navigation needs and preferences of wheelchair users were investigated. MAGUS [12] is a web-based navigation service that provides pre-trip planning for wheelchair users in urban areas. Optimal routes are computed with respect to six routing criteria: shortest distance, minimum barriers, fewest slopes, avoiding challenging surfaces, using only controlled crossings, and limited road crossings. U-Access [13] is a web-based routing tool that provides shortest routes to given destinations for people with three ability levels: unaided mobility, aided mobility (using crutch, cane or walker) and wheelchair users. The evaluation results showed that the route from an origin to a destination varies for each ability level by the distance travelled and the difficulty of the routes. Karimanzira et al. [14] developed a travel aid to assist the visual/ limb/hearing impaired for pre-trip planning in urban areas and tested it in Georgenthal, Germany. Law of physics and an energy cost mathematical model were used to determine the effort required to overcome barriers on sidewalk segments. Ourway [15] is a mobile pedestrian navigation prototype with special features for people with mobility impairments or parents with baby strollers. The prototype provides the users with a tool for rating the path and employs those ratings to recommend routes to other users. RouteCheckr [16] presents a multimodal annotation prototype that allows users to annotate existing geographical data with their own information such as safety rate. This additional information would be useful for optimal route computation. However, it requires a large amount of user’s collaboration to share experience on travelled routes before suitable routes can be recommended. Clearly, current research projects do not fully address all the requirements of wheelchair users for navigation and there is a need for wheelchair navigation systems that consider environmental barriers and user’s aversions to certain obstacles. In our previous work [17], we identified environmental obstacles that have an impact on accessibility and safety of wheelchair users and developed a personalized routing technique that includes three methods to weigh the impedance score of each sidewalk segment for individual wheelchair users. The calculated impedance scores of sidewalk segments in the network are then used to compute personalized routes between given pairs of origins and destinations. The three weighting methods were tested using simulated preferences of wheelchair users and it was found that the Absolute Restriction Method (ARM) is more effective for generating personalized routes than the other two weighting methods. However, the personalized routing technique based on ARM has never been experimented with wheelchair users. The purposes of the work presented in this article are to validate the personalized routing technique with wheelchair users, to understand wheelchair users’ route choices and to acquire wheelchair users’ feedbacks on the necessity of wheelchair navigation systems/services and the importance of personalized routes. We developed a web-based system that provides personalized routes for wheelchair users based on the ARM and evaluated it with five manual wheelchair users. The remainder of the article is structured as follows. The section titled ‘‘Methods’’ provides details of methods used in the study. The section titled ‘‘Experimental results’’ discusses the experimental results and the section titled ‘‘Discussion’’ provides the discussion. Conclusions and future research

199

are provided in the section titled ‘‘Conclusions and future research’’.

Methods Personalized routing for wheelchair navigation Routing is an important component in any navigation system/ service which computes the best route between two locations given user’s preferences [18]. For wheelchair navigation, routing requires a specific digital map database that represents the geometry and topology of the network segments as well as attributes information about those segments. In this article, seven parameters, listed as important outdoor obstacles, are: width, length, slope, surface type, surface conditions and traffic. These seven parameters are chosen based on the review and analysis of data requirements for wheelchair navigation [19], the review of course materials on obstacles to assess wheelchair skills [4], the pedestrian access guidelines [20] and the feasibility of collecting the required data. The pedestrian access guidelines published by Federal Highway Administration (FHWA) describe four critical sidewalk characteristics [20] as follows. First, the sidewalks should have at least 3600 clearance width and not be blocked with obstacles (e.g. poles and signs). Second, sidewalk grade and cross-slope should not exceed 5% and 2%, respectively, in order to be safe for wheelchair users. Third, walking surface must be firm, stable and slip resistant. Generally surface materials are made up of concrete, asphalt, stone and brick. Ideally, sidewalks should not be broken or have cracks. Lastly, curb ramps should be provided, especially at intersections, in order to provide access for wheelchair users between a sidewalk and a street. The definition, data collection method and values for each parameter are summarized in Table 1. Personalized routing composes of two steps: route preference quantification and route calculation. Route preference quantification determines a weight indicating user’s preference for each segment. Weighted segments are then used in the route calculation step where an algorithm considers each segment’s weight in order to determine an optimal route. In our weighting method, called ARM [17], the route preference quantification step ranks the parameters (Table 1) of a pedestrian network according to user’s preferences using the Analytic Hierarchy Process (AHP). The rank of each parameter determines its importance in terms of tendency to impede travel, as indicated by the user. Once each parameter is weighted, the impedance level score of each segment can be determined by aggregating the weighs of all parameters with their corresponding segment’s attributes values through the fuzzy logic technique. The impedance level score for each segment may vary for different users depending on their preferences. Higher impedance level scores indicate more difficulty while traveling along the segment. The scores range from 0.01 to 5, where 0.01 is the least difficult and 5 is the most difficult. A score of 0 is assigned to segments that are impassable and thus cannot be rated. The last task of the route preference quantification step is to assign a cost to each pedestrian path segment, calculated by Equation (1).

Cost ¼ Impedance level  ðLWF  lengthÞ  Width  Steps ð1Þ The cost value is obtained by multiplying the impedance level with segment length, its weigh factor (LWF), segment’s width and steps. The cost value is 0, where 0 means that there is an impassable path. The higher the cost, the greater the level of impedance will be. As shown in Table 1, a pedestrian path segment with width and steps that are not compliant with the ADA standards (a width of 50.914 m and a segment with one or

200

P. Kasemsuppakorn et al.

Disabil Rehabil Assist Technol, 2015; 10(3): 198–210

Table 1. Summary of the parameters and their values for pedestrian networks. Parameters

Data collection method

Width

Clearance width of sidewalk

Field survey

Length

Geometric distance of a pedestrian path segment The steepness, incline or grade on the pedestrian path segment

Step

An abrupt change in level 40.01 m

Using GIS tools to calculate geometric distance from end points and shape points Using GIS tools (Spatial statistics) to calculate degree of slope from digital elevation model (DEM) data Field survey

Surface type

Material of surface type

Field survey

Surface condition

Cracks, manhole cover and uneven surface

Traffic

Passage of people along pedestrian path segments that depend on day and time

– Field survey recording obstacles’ locations and sizes using GPS units – Quantify and calculate the surface condition score from combination of cracks, manhole cover, uneven surface Categorize sidewalk and determine traffic level based on day and time

Slope

Downloaded by [Australian National University] at 15:51 05 November 2015

Description

more steps) is defined as impassable path (cost ¼ 0) by assigning width attribute value and steps attribute value to 0. For details of the route preference quantification step in ARM refer to Kasemsuppakorn and Karimi [17]. The route calculation step computes optimal routes, employing Dijkstra’s algorithm [21], to find the least impedance route from a pair of origin and destination. The input to this step is the weighted pedestrian network in a given area and the coordinates of a valid origin-destination pair. Evaluation The effectiveness of our personalized routing technique was validated through an experiment using a web tool. The experiment was conducted within the University of Pittsburgh’s main campus, which features environmental challenges (e.g. hilly terrain) for wheelchair users. The pedestrian network database used in this experiment was the updated version of the database developed by Kasemsuppakorn and Karimi [17]. This pedestrian network was manually digitized by using high-resolution imagery and GIS tools. A field survey, using both paper maps and Global Positioning System (GPS) data collection, was performed in the University of Pittsburgh’s main campus to verify the collected pedestrian network and to collect campus buildings’ accessible entrances. Inclusion/exclusion criteria The evaluation study was approved by the Institutional Review Board (IRB) at the University of Pittsburgh. Participants were included if they were over 18 years of age, used a manual wheelchair as their primary means of mobility (420 h/week), and could independently operate a manual wheelchair without assistance. Wheelchair users who took part in the study provided a written informed consent prior to their participation in the study. Experimental protocol The protocol consisted of three sessions: (1) pre-activity questionnaire session, (2) activity session and (3) post-activity questionnaire session. During the pre-activity session, participants

Values 0–50.914 m 1–0.914 m Distance (m) Degree of slope (%) 0 – Number of steps 40 1 – Number of steps ¼0 1 – Concrete 2 – Asphalt 3 – Brick 4 – Cobblestone 5 – Gravel Numerical value

1 – No congestion 2 – Light traffic 3 – Heavy traffic

were asked to complete a pre-activity questionnaire (Appendix A), that includes questions on demographics, wheelchair information, type of disability and perceived level of fitness. The level of fitness was determined by using a scale from 1 to 10, where 1 indicates a low fitness level (e.g. get tired easily and avoid inclines) and 10 indicates a high fitness level (e.g. take a lot to get tired and can travel steeper slopes). They were also asked to compare different pedestrian path segment parameters including slope, steps, distance, sidewalk width, sidewalk surface, sidewalk traffic against each other and rate their perceived difficulty levels for each pair of parameters by assigning weights to parameters through the web-based interface. The weighted parameters were then used to calculate personalized routes. An example of the preference scale (left) and the calculated weight value of each parameter (right) is shown in Figure 1. During the activity session, subjects were asked to travel to a destination, a building, within the University of Pittsburgh’s campus. Each subject traveled from the same origin building (Forbes Tower), to different destination buildings. The destination buildings were selected by the investigators rather than the participants due to the lack of their familiarity with the campus area. Based on the subject ratings on their fitness levels and preferences for different sidewalk parameters, and the destination building identified, the routing algorithm mentioned in the previous section calculates a personalized route. For comparison purpose, another route with the shortest feasible distance to the same destination was generated. This process took 55 min. During the process, a GPS datalogger (Ohararp, LLC) was attached to the backrest of subject’s wheelchair for recording the traveling information along the routes such as distance, speed and GPS coordinates. After the two routes were generated, participants were instructed to follow two different routes from start to finish with a return trip to start. They traveled with their own wheelchairs and were accompanied by at least two investigators. One provided directions to the subjects while the other investigator documented the number of assistance required, and provided assistance, when needed. The order of the routes was randomized. Subjects did not know which route was the personalized one. Example routes between a pair of campus buildings are shown in Figure 2.

Route choices for wheelchair navigation

Downloaded by [Australian National University] at 15:51 05 November 2015

DOI: 10.3109/17483107.2014.898160

201

Figure 1. An example of the preference scale (prioritizing parameters).

Figure 2. Personalized route and shortest route between a pair of campus buildings.

During the post-activity questionnaire session, subjects were asked to complete a questionnaire (Appendix B) that included questions on their ratings for the parameters of the two routes (e.g. width, slope, surface) and their comfort on a visual analog scale

where they placed a vertical mark on a 10-cm line to indicate poor (0 cm) to excellent ratings (10 cm) for the features. They also rated the overall effort when traveling along each route using a visual analog scale where a rating of zero indicated no effort and a rating

202

P. Kasemsuppakorn et al.

Disabil Rehabil Assist Technol, 2015; 10(3): 198–210

Table 2. The characteristics of participants. Subject ID 1 2 3 4 5

Age

Gender

Disability type

Years of wheelchair use

Wheelchair make and model

Perceived fitness level

Most concern parameter

23 28 29 39 26

Female Male Male Male Male

Orthopedic impairments Spinal cord injury Spinal cord injury Spinal cord injury Spinal cord injury

4 24 10 20 17

TiLite XC TiLite X Invacare Terminator Titanium TiLite TR Quickie Ti

2 9 8 10 10

Slope Width Surface condition Surface condition Surface condition

Table 3. Traveling information during the activity session. Route characteristics

Downloaded by [Australian National University] at 15:51 05 November 2015

Subject ID

Destination building

1

Mervis Hall

2

Thackery Hall

3

Wesley W. Posvar Hall

4

Pittsburgh Athletic Association

5

Frick Fine Art Building

Routes

Route distance traveled (m)

P* NP* P* NP* P* NP* P* NP* P* NP*

1160.12 917.24 1291.46 1160.04 1237.10 1017.24 1755.92 1568.66 1845.02 1559.08

Range of slope values [Min–Max] (degree) [0.88–4.46] [1.27–5.54] [0.87–4.85] [0.87–6.35] [0.81–4.84] [0.88–3.73] [0.88–4.06] [0.89–5.44] [0.78–8.90] [0.87–7.99]

average: average: average: average: average: average: average: average: average: average:

1.49 1.77 1.58 2.97 1.45 1.62 1.43 2.98 1.06 1.85

Average surface condition score

Travel time (min)

Average travel speed (m/s)

2.63 3.31 0.28 1.40 2.44 3.31 0.96 2.33 1.96 2.83

33 41 19 18 16 13 23 20 17 15

1.57 1.48 2.46 2.40 2.14 2.23 2.01 1.93 2.05 2.09

P* indicates personalized route and NP* indicates shortest feasible route.

of 10 indicated maximum effort. In addition, they were asked to provide any suggestions that could help develop a personalized routing system. Data collection and analysis Descriptive statistics were used for all variables collected in the questionnaires at the beginning and end of the study. Traveling information during the activity session was also summarized based on the GPS data logger. Due to the small sample size, no statistical analysis was performed. The experimental results and discussion are described in the sections titled ‘‘Experimental results’’ and ‘‘Discussion’’, respectively.

Experimental results Through a pre-activity questionnaire, the demographics and wheelchair information of five participants were collected as shown in Table 2. All participants were Caucasian and used ultralight manual wheelchairs. Four male and one female wheelchair users, who were 23–39 years of age and were unfamiliar with the study area, participated in this study. All subjects were experienced manual wheelchair users (maximum years of wheelchair use were 24 and minimum was 4). Four of them had spinal cord injury and one had orthopedic impairment. The lowest perceived level of fitness was 2 (subject 1) and the highest perceived level of fitness was 10 (subjects 4 and 5). Answers from the questionnaire revealed that three subjects (subjects 3, 4 and 5) were concerned with the surface condition (e.g. surface type, cracks, unevenness), subject 1 was concerned with slope and subject 2 was concerned with the sidewalk width. During the activity session, data collected include total travel time and average speed of each subject along each route. The travel time included the time spent operating the wheelchair and the time spent resting before reaching the destination (round trip). In addition, we also calculated such quantities as total traveled distance (round trip), minimum, maximum and average of path’s slope in degree, and average surface condition score along each route for comparison, as shown in Table 3.

Table 3 is discussed through two examples as follows. Subject 1 traveled from Forbes Tower to Mervis Hall, with the average speed of 1.57 m/s, in 33 min along the roundtrip personalized route and with the average speed of 1.48 m/s, in 41 min along the roundtrip shortest feasible route. In regard to route characteristics for this subject, the personalized route (P*) had longer distance (1160.124917.24) but less maximum degree of slope along the path (4.4655.54) and better surface condition (2.6353.31) than the shortest feasible route (NP*). Another example is subject 3 who traveled from Forbes Tower to Wesley W. Posvar Hall. The subject traveled with the average speed of 2.14 m/s, in 16 min along the roundtrip personalized route and with the average speed of 2.23 m/s, in 13 min along the roundtrip shortest feasible route. The personalized route had longer distance (1237.1041017.24) and contained some steeper segments (4.8443.73), but better surface condition (2.4453.31) than the shortest feasible route. In the post-activity session, subjects provided their own ratings on each parameter and their perceived overall comfort and effort when traveling along the two routes, as illustrated in Figure 3. The sidewalk parameters were rated on a 0–10 scale with higher scores indicating greater satisfaction with the parameter. Four out of five subjects rated the parameters (highlighted with the ellipse in Figure 3) they deemed most important for personalized routes compared with shortest feasible routes. The only exception was subject 2 who chose width as the most important parameter and rated the personalized route width lower than the shortest feasible route. Subject 1 rated slope and surface condition for personalized routes higher than for shortest feasible routes, whereas the pedestrian traffic parameter was rated lower for personalized routes and higher for shortest feasible routes. The width and distance parameters were rated the same for both routes. Subject 2 also rated slope and surface condition along personalized routes higher than shortest feasible routes. However, this subject rated width, distance and traffic along shortest feasible routes higher than personalized routes. Subject 3 indicated higher satisfaction with surface condition and width along personalized routes than along shortest feasible routes, whereas slope, distance and

Route choices for wheelchair navigation

Downloaded by [Australian National University] at 15:51 05 November 2015

DOI: 10.3109/17483107.2014.898160

203

Figure 3. The rating comparison between two types of routes by five participants.

pedestrian traffic parameters were rated lower for personalized routes. Subject 4 rated slope, surface condition and width for personalized routes higher than shortest feasible routes, while the distance parameter of the personalized route was rated lower than the shortest feasible route. The pedestrian traffic parameter was rated the same for both routes. Subject 5 had a greater satisfaction with slope, width and distance along personalized routes than shortest feasible routes, whereas slope and pedestrian traffic were rated lower for personalized routes. The ratings for the perceived overall comfort and effort also ranged from 0 to 10. For the perceived overall comfort, higher scores indicate a higher comfort level. On the other hand, for the perceived effort, higher scores indicate higher required effort. Subjects 1, 2 and 4 stated that the personalized routes required less effort to travel and provided more comfortable paths than the shortest feasible routes. Nevertheless, subjects 3 and 5 indicated that the personalized routes required more effort to travel and provided less comfortable paths than the shortest feasible routes.

The section titled ‘‘Methods’’ of the post-activity questionnaire asked subjects to provide their opinion on and suggestion for personalized wheelchair navigation. All five subjects showed interest in utilizing real-time personalized wheelchair navigation devices such as a cell-phone or a PDA. Subjects also suggested some additional features including information about routes in indoors (e.g. connected buildings), information about sidewalk traffic, and a touch screen. They also indicated the inclusion of other parameters, such as cross slopes, type of crosswalks, sprinklers and snow piles, for travel planning. Cross slope is considered by FHWA as an important parameter for accessible routes. However, we did not consider cross slope in our study for the reasons that they are not available in existing databases and that their measurement involves complex procedures.

Discussion The participants’ demographics and wheelchair information showed that they were fairly homogeneous with respect to their

Downloaded by [Australian National University] at 15:51 05 November 2015

204

P. Kasemsuppakorn et al.

age, experience using a wheelchair, wheelchair model, unfamiliarity with the study area, and disability types. Four of the five subjects had a spinal cord injury, one had an orthopedic problem, and none of the subjects had cognitive impairment. All subjects had good health and their disabilities did not significantly impede their physical activities since they were able to complete the experiment and travel with at least the average speed of manual wheelchair users (0.96 ± 0.17 m/s [22]). Subject 1 traveled a shorter distance and took a longer amount of time than the other subjects due to a low fitness level and lack of prior wheelchair usage. Overall, the distance the subjects travelled was 14.64% greater for the personalized routes than the shortest feasible routes. However, all personalized routes had a better surface condition than the shortest feasible routes. Nevertheless, subjects 3 and 5 encountered some steeper segments (see maximum slope) along the personalized route which may explain the longer travel times as compared to the shortest feasible route times and the travel times of the other subjects. Most subjects (except subject 1) spent more time traveling the personalized route as opposed to the shortest feasible route due to the longer distance of the personalized route. Subject 1 had a long resting time along the shortest feasible route resulting in a longer total travel time. As reported in the ‘‘Experimental result’’ section, overall the rating comparison between the personalized routes and the shortest feasible routes showed that most subjects indicated a higher rating for the personalized route in terms of slope, surface condition and width. However, in terms of traveling distance and traffic, the shortest route received a better rating. Specifically, the most dominant parameter preventing wheelchair mobility is surface condition (material of surface type and the quality of the surface) since three out of five subjects weighted this parameter the highest and the remaining two subjects weighted it the second highest. A comparison of the rating results reveals that the personalized route’s surface condition was more satisfactory than the shortest feasible route. Therefore, in this case, the personalized route reflected the needs of wheelchair users to a greater degree than the shortest feasible route. Moreover, most subjects rated their most important parameters higher for the personalized route than for the shortest feasible route. The only exception was subject 2 who chose ‘‘width’’ as an important parameter and rated it higher for the shortest feasible route. In this case, the dissatisfaction with the width of the personalized route might be explained by obstructions such as mobile advertisement signs placed on the sidewalk in front of the stores along the personalized route. This leads us to suggest that signs on sidewalks, among similar obstructions, should be considered as another parameter to calculate routes for wheelchair users. Even though sidewalk signs do not cause safety concerns, they might block a path requiring detours along routes. Comfort and effort have an inverse relationship where higher comfort values led to lower effort levels and vice versa. Three out of five subjects indicated that the personalized routes were more comfortable than the shortest feasible routes. The same three subjects also reported that they spent less effort traveling along the personalized routes than the shortest routes. Two subjects (subjects 3 and 5) had negative feedback on the degree of comfortableness and effort. This is because both subjects encountered some segments with higher degree of slope along the personalized routes than the shortest feasible routes. Moreover, subject 3 travel took place during the period when the traffic was high and the personalized route for subject 5 consisted of more uncontrolled crosswalks (no pedestrian crossing signal) than the shortest route; uncontrolled crosswalks led to segments with high traffic and required more effort to

Disabil Rehabil Assist Technol, 2015; 10(3): 198–210

travel. However, number and type of crosswalks were not considered in the study and should be included in future research. This study was primarily limited by its small sample size and homogeneity of the participants. Therefore, it does not represent the needs of all types of wheelchair users and limits the generalizability of the research. Although personalized wheelchair navigation is a promising solution to lessen the difficulties that manual wheelchair users face when navigating unfamiliar environments, it requires a complete pedestrian network containing geometry and topology along with attribute information as described in Table 1. In the case of unavailable data, manual digitization and field survey may be employed to generate such data for a given area. For other map acquisition approaches refer to [23]. Since an appropriate pedestrian network is important in wheelchair navigation services, other approaches such as the Volunteered Geographic Information (VGI) [24] by wheelchair users in the area could be used to complement the collection of the required data. Also urban planning is another solution that could improve the accessibility of the community and increase the quality of life for individuals with mobility impairments.

Conclusions and future research Understanding route choices for wheelchair navigation is a major task in developing navigation systems/services that can address the specific needs and preferences of wheelchair users. In this article, the Absolute Restriction Method (ARM) technique that combines individual variables, route preferences and environmental characteristics to calculate personalized routes for wheelchair users was tested. The five subjects participating in the study were homogeneous with respect to their age, experience using a wheelchair, wheelchair model, unfamiliarity with the study area and disability types. From the pre-activity questionnaire, the most dominant parameter preventing wheelchair mobility is surface condition (material of surface type and the quality of the surface). According to the rating comparison between personalized routes and shortest feasible routes, four out of five subjects rated the parameters they deemed most important higher for the personalized routes than the shortest feasible routes and three of them indicated overall that personalized routes were more comfortable than shortest routes. From the feedbacks, other parameters need to be considered in route calculation such as sidewalk signs, maximum degree of slope along the path, number of crosswalks and crosswalk types. We believe that this study could also help those who use baby strollers, cane, crutch and walker to plan their visits to the community. In our future research we plan to: (1) expand our experimentation by using a larger number of subjects with more diversity; (2) find a pattern of preferred routes for groups of users with similar characteristics; (3) include all of the required environmental parameters with recommended values from FHWA (e.g. cross slope) and the suggested parameters by the subjects; (4) include personal variables related to the destination (e.g. level of importance of the destination) in the routing function and (5) augment our routing function with route annotation from wheelchair users in the community.

Acknowledgements The authors would like to thank Ms. Jessica Benner for her assistance during the subject testing part of the project and for providing feedback on the early drafts of the manuscript.

Declaration of interest The authors report no declarations of interest.

DOI: 10.3109/17483107.2014.898160

Downloaded by [Australian National University] at 15:51 05 November 2015

References 1. Rimmer JH, Riley B, Wang E, et al. Physical activity participation among persons with disabilities barriers and facilitators. Am J Prev Med 2004;26:419–25. 2. Ding D, Parmanto B, Karimi HA, et al. Design considerations for a personalized wheelchair navigation system. The 29th Annual International Conference of the Engineering in Medicine and Biology Society (EMBS). Lyon, France: IEEE; 2007:4790–3. 3. Meyers A, Anderson J, Miller D, et al. Barriers, facilitators, and access for wheelchair users: substantive and methodologic lessons from a pilot study of environmental effects. Soc Sci Med 2002;55: 1435–46. 4. Kilkens OJ, Post MW, Dallmeijer AJ, et al. Wheelchair skills tests: a systematic review. Clin Rehabil 2003;17:418–30. 5. Volkel T, Kuhn R, Weber G. Mobility impaired pedestrians are not cars: requirements for the annotation of geographical data. In: Miesenberger K, Klaus J, Zagler W, Karshmer A, eds. The 11th International Conference, ICCHP. Linz, Austria: Springer; 2008:1085–92. 6. Board USA. Americans with disabilities act and architectural barriers act accessibility guidelines; 2004. Available from: http:// www.wbdg.org/ccb/ASTAND/ada_aba.pdf [last accessed August 2013]. 7. Yano Research. World Car Navigation & PND Market: Research Findings 2010. Research Summary 2011. Available from: www.yanoresearch.com/press/pdf/730.pdf [last accessed July 2013]. 8. Patel K, Chen MY, Smith I, Landay JA, eds. Personalizing Routes. The 19th annual ACM symposium on User interface software and technology (UIST). Montreux, Switzerland; 2006:187–90. 9. Butcher D. Mobile Marketer. 2009. Available from: http://www.mobilemarketer.com/cms/news/media/3397.html [last accessed July 2013]. 10. Finocchiaro P. Mobile Marketer. 2010. Available from: http:// www.mobilemarketer.com/cms/news/search/7132.html [last accessed July 2013]. 11. Karimi HA, Jiang M, Zhu R. Pedestrian navigation services: challenges and current trends. Geomatica 2013;67:259–71. 12. Beale L, Field K, Briggs D, et al. Mapping for wheelchair users: route navigation in urban spaces. Cartographic J 2006;43:68–81.

Route choices for wheelchair navigation

205

13. Sobek AD, Miller HJ. U-Access: a web-based system for routing pedestrians of differing abilities. J Geogr Syst 2006;8: 269–87. 14. Karimanzira D, Otto P, Wernstedt J. Application of machine learning methods to route planning and navigation for disabled people. The 25th IASTED International Conference Lanzarote. Canary islands, Spain; 2006:366–71. 15. Holone H, Misund G, Holmstedt H, eds. Users are doing it for themselves: pedestrian navigation with user generated content. Internation Conference On Next Generation Mobile Applications, Services and Technologies (NGMAST ’07). Cardiff, UK: IEEE; 2007:91–9. 16. Volker T, Weber G. RouteCheckr: personalized multicriteria routing for mobility impaired pedestrians. The 10th international ACM SIGACCESS conference on Computers and accessibility. Halifax, Nova Scotia, Canada: ACM; 2008:185–92. 17. Kasemsuppakorn P, Karimi HA. Personalised routing for wheelchair navigation. J Loc Based Serv 2009;3:24–54. 18. Karimi HA, Roongpiboonsopit D, Kasemsuppakorn P. Uncertainty in personal navigation services. J Navigat 2011; 64:341–56. 19. Kasemsuppakorn P, Karimi HA, Data requirements and a spatial database for personalized wheelchair navigation. The 2nd International Convention on Rehabilitation Engineering & Assistive Technology. Bangkok, Thailand; 2008:31–4. 20. Kirschbaum JB, Axelson PW, Longmuir PE, et al. Designing sidewalks and trails for access: Part II of II: best practices design guide. Publication No. FHWA-EP-01-027. Washington, DC: US Department of Transportation, Federal Highway Administration; 2001. 21. Dijkstra EW. A note on two problems in connexion with graphs. Numer Math 1959;1:271. 22. Tolerico ML, Ding D, Cooper RA, et al. Assessing mobility characteristics and activity levels of manual wheelchair users. J Rehabil Res Dev 2007;44:561–72. 23. Karimi HA, Kasemsuppakorn P. Pedestrian network map generation approaches and recommendation. Int J Geogr Inform Sci 2012;27: 947–62. 24. Goodchild MF. Citizens as sensors: the world of volunteered geography. GeoJournal 2007;69:211–21.

206

P. Kasemsuppakorn et al.

Disabil Rehabil Assist Technol, 2015; 10(3): 198–210

Appendices Appendix A Pre-Activity Questionnaire Please take a few minutes to answer the following questions. Your honest opinions will be very helpful to improve the effectiveness of the personalized route generation methodology. Section 1: Background information and health status Date: ___/___/______ Name: ________________________________________________________________________ Address: ______________________________________________________________________

Downloaded by [Australian National University] at 15:51 05 November 2015

City :_________________________________State: ___________________________________ Zip code:______________________________________________________________________ Telephone number:____________________________Email: ____________________________ gender

Male

Female

Date of Birth: ___/___/______ Age: ______ Years Ethnic Origin: African American Asian American Caucasian Hispanic Native American Other ___________________ Manual Wheelchair Make (Brand): Action/Invacare

Permobil

Everest and Jennings

Pride

Kuschall

Sunrise/ Quickie

Otto Bock

TiLite/ TiSport

Other (Please specific):____________________________________________________ Manual Wheelchair Model:_______________________________________________________ When did you start using a manual wheelchair: ___/___/______ What type of disability do you have? Spinal Cord Injury, Level______________ Stroke Cerebral Palsy Multiple Sclerosis Traumatic Brain Injury Amputee Other, Specify ______________________

Route choices for wheelchair navigation

DOI: 10.3109/17483107.2014.898160

What is your level of fitness? (Please circle the numbers)

1

2

3

4

5

6

7

8

9

10

1 : I get tired very easily and avoid inclines. 10: It takes a lot to make me tired and I can travel steeper slopes without a problem.

What is the farthest distance that you can go? About one or two city blocks About one mile

Section 2: Ranking the sidewalk parameters. Instructions: This part of the questionnaire requests you to compare each parameter with one and other. In the table below, each row compares the parameter in the far left column with the parameter in the far right column in terms of how much they prevent you from getting to where you want to go. Please select which parameter is more of a problem for travel than the other by checking the appropriate box in between the parameters being compared. Only select one box per row.

The meaning of each parameter Slope: The steepness of the sidewalk parallel to the direction of travel. Steps: Change in level greater than 13mm (ADA), without a ramped transition. Width: Sidewalk width Distance: Sidewalk distance Sidewalk surface: Material on the pedestrian environment such as concrete, asphalt, as well as sidewalk condition such as cracks, uneven surface Sidewalk Traffic: Density of users traveling on the sidewalk.

Extremely

Very Strongly

Strongly

Moderately

No Difference

Moderately

Strongly

Parameter

Very Strongly

Example:

Extremely

Downloaded by [Australian National University] at 15:51 05 November 2015

More than one mile

Parameter

Steps

Slope

Steps is very strongly important

X than slope. Slope

Width

Slope is extremely important than

X sidewalk width.

207

208

P. Kasemsuppakorn et al.

Disabil Rehabil Assist Technol, 2015; 10(3): 198–210

Downloaded by [Australian National University] at 15:51 05 November 2015

Extremely

Very Strongly

Strongly

Moderately

No Difference

Moderately

Strongly

Very Strongly

Parameter

Extremely

Which parameter prevents your mobility more than the other?

Parameter

Slope

Steps

Slope

Width

Slope

Distance

Slope

Sidewalk surface

Slope

Sidewalk Traffic

Steps

Width

Steps

Distance

Steps

Sidewalk Surface

Steps

Sidewalk Traffic

Width

Distance

Width

Sidewalk Surface

Width

Sidewalk Traffic

Distance

Sidewalk Surface

Distance

Sidewalk Traffic

Sidewalk Surface

Sidewalk Traffic

Route choices for wheelchair navigation

DOI: 10.3109/17483107.2014.898160

Appendix B Post-Activity Questionnaire

The post-activity questionnaire will be complete after the activity session. Please take a few minutes to answer the following questions. Your honest opinions will be very helpful to improve the effectiveness of the personalized route generation system. Section 1: Please place a vertical mark on the line below to indicate your opinion on each criterion of both routes. Route 1: the first route that you traveled

Downloaded by [Australian National University] at 15:51 05 November 2015

Route 2: the second route that you traveled

1. What do you think about the slope along the route?

Route 1:

Poor

Excellent

Poor

Excellent

Route 2: 2. What do you think about the sidewalk surface and sidewalk condition such as crack?

Route 1:

Poor

Excellent

Poor

Excellent

Route 2: 3. What do you think about the sidewalk width for the overall paths?

Poor

Excellent

Poor

Excellent

Route 1:

Route 2: 4. What do you think about the steps or stairs along the route?

Poor

Excellent

Poor

Excellent

Route 1:

Route 2: 5. What do you think about the total distance along the route?

Poor

Excellent

Poor

Excellent

Route 1:

Route 2:

209

210

P. Kasemsuppakorn et al.

Disabil Rehabil Assist Technol, 2015; 10(3): 198–210

6. What do you think about the sidewalk traffic along the route?

Poor

Excellent

Poor

Excellent

Route 1:

Route 2: 7. How comfortable is your traveling?

Poor

Excellent

Poor

Excellent

Route 1:

Downloaded by [Australian National University] at 15:51 05 November 2015

Route 2:

8. How much effort do you feel when traveling along the routes?

A little

Very

A little

Very

Route 1:

Route 2:

Section 2: Please give us your opinion and suggestion 1. Are there any other sidewalk parameters that are the obstacles for traveling?

2. If there is a real-time personalized wheelchair navigation device based on a cell-phone or a PDA, would you like to use it?

Yes (go to 2.1)

No (please give us the reason)

Reason _____________________________________________________________

2.1 Where is an appropriate place to mount it? ________________________________________________________________ 2.2 What features you would like to see in this real-time personalized wheelchair navigation device? ___________________________________________________________________________ ___________________________________________________________________________ _________________________________________________________

Understanding route choices for wheelchair navigation.

To validate a personalized routing technique with wheelchair users, understand their route choices and acquire their feedback on the necessity of whee...
2MB Sizes 2 Downloads 3 Views