Davidmcgookin.files.wordpress.com

Initial Development of a PDA Mobility Aid for David McGookin1, Maya Gibbs1, Annu-Maaria Nivala2, and Stephen Brewster1 1 Department of Computing Science, University of Glasgow, Glasgow G12 8QQ, UK {mcgookdk,stephen}@dcs.gla.ac.uk 2 Finnish Geodetic Institute, Geodeetinrinne 2, P.O. Box 15, Finland Abstract. We discuss requirements surrounding a mobile navigationsystem for visually impaired people. We describe an initial prototypebased on a PDA using GPS location tracking. This prototype has so farfailed to provide reliable location detection, due to the use of GPS in builtup environments. We discuss how our system may improve detection byswitching between a range of different location tracking technologies.
However, we conclude that there may still be times when these technolo-gies fail, and more work is needed on how to support the user in suchcircumstances.
When navigating an environment, visually impaired people are presented witha number of problems. Hazard avoidance such as street furniture and unevensurfaces can be overcome with the use of a white cane or guide dog. However“macro-navigation”, finding directions from A to B [1], presents more problems.
Many local government organisations provide mobility officers who can help avisually impaired person navigate the environment, and help with learning (orre-learning) locations. However, such services are limited, and not of use forpeople visiting other towns or places when on business or holidays. A way toovercome these problems would be the use of a mobile navigation device, whichwould help to guide a visually impaired person around an environment. Severalsuch systems have been proposed, but few evaluated. The most notable wasMoBIC developed by Petrie et al. [1]. Further research has been carried outsince then, but mostly on different tracking systems and alternate interfaces todisplay information to the user [2]. Whilst these are important, in the last fewyears the growth in availability and power of mobile devices, as well as increasingmobile availability of the internet, has increased the potential functionality ofmobile navigation systems for the visually impaired. We believe it is thereforeappropriate to investigate mobile navigation systems on low cost mobile devices.
In this paper we present our initial requirements capture and prototype system,as well as issues for future design decisions.
To further understand the problems of navigation for visually impaired people,and determine requirements for our system, we carried out a semi-structuredinterview with a middle aged, blind individual who worked as a transcriptionofficer for the Royal National Institute of the Blind (RNIB) in the UK. Theinterview covered navigation in the world and how this was accomplished.
An emphasis on macro rather than micro navigation was emphasised. The use of a white cane to detect immediate obstacles was successful, but the abilityto navigate to and from unfamiliar locations was more difficult. As the partic-ipant noted, “I certainly wouldn’t turn up at, say, Aberdeen (railway) stationand expect to navigate my way somewhere. Even if it is only within 200 yards.”.
The participant also noted than in addition to just directions from A to B, in-formation about temporary impediments to navigation would be useful. Theseseem to fall somewhat between micro and macro navigation, being things whichwould be detectable with a white cane aid, but would be beneficial to avoidall together. For example, when a pavement (sidewalk) is being resurfaced andpedestrians are diverted through a sectioned off area of the road. Additionally,he noted that it would be useful to select routes on more that simply short-est distance, with other factors such as avoiding traffic or steep inclines beingpreferred. The participant felt that people “don’t generally like looking out ofplace, or sounding out of place”, so the system should not draw attention to theuser. Therefore speech input was felt to be inappropriate, but speech output waspreferred as walking using headphones “doesn’t look too out of place these days”.
The requirements identified here are broadly in line with those identified by Strothotte et al. [3] as part of to MoBIC project, highlighting that users stillhave the same basic needs although technology has moved on in the last decade.
Based on the requirements identified, we designed and implemented an initialprototype to provide route navigation abilities. Once the initial capabilities havebeen implemented and evaluated, we will be able to implement more of thefeatures identified during our requirements capture.
Our initial system runs as an application on a Dell Axim Pocket PC, and retrieves location information using the global positioning system (GPS) via aTomTom Bluetooth GPS receiver. At the moment the application automaticallydetects the user’s location and then calculates a route to a destination based onselection from a list. We store data in files formatted in Geographic Mark-upLanguage which is a standard XML compliant format developed for cartographicinformation. Once loaded into our system the map is held as a weighted graph(each road segment is assigned a “weight” dependent on its length), and theroute to the user’s destination is calculated using a shortest path algorithm.
That is to say that the application calculates the route with the lowest sumtotal of weights. Whilst in our initial system weightings are strictly geographic distances, we can easily change the weightings to reflect different navigationpriorities, such as avoiding particularly steep or busy roads. In the future wemay be able to interface the system, through wireless networks, to local councilworks notifications to bias against areas where road works or other constructionis occurring, thereby addressing some of the issues identified in our interview.
As was requested in the requirements phase, synthetic speech was used to provide all feedback from the system. Once the user has selected a destination,the system informs what direction to walk in, and then when to turn left orright at appropriate intersections. The user can also request the system providesthe current location. Output is played through a single ear piece or, if the userprefers, the PDA’s in-built speaker.
To simplify user input we have applied a simple menu based interface. A back, forward and select button are presented on the PDA screen and can befelt by the user via a raised paper control panel overlaying the touch screen. Thisallows controls to be shaped differently and therefore be easier to identify thanusing the inbuilt buttons on the PDA, or using the keypad of a smart phonedevice. However we must still identify how well this interface will scale whenmore functions are added.
So far we have not carried out any field trials of our system using visuallyimpaired people. We have however carried out technical evaluations to establishthe quality of information available from our system, and a short user evaluationof the interface. Whilst users found the interface straightforward to navigate, thelocation detection of our system has been less successful. Our system, in a desireto use “off the shelf” location tracking, uses standard GPS. In our tests thishas proven to be an inadequate technology to support the navigation tasks,with many occasions where the system failed to correctly detect its positionand inform the user when to change direction. That GPS is poor in built upenvironments is a known problem (our test area was on a hill surrounded by tallbuildings), and many others have proposed using differential GPS [2] which usesan additional ground transmitter to improve location. However these systemsstill require line of sight with GPS satellites, otherwise incorrect locations willbe determined. For example, the MoBIC system was evaluated in an area withopen spaces [1]. However many towns and cities do not always afford such openspace. Other systems propose that the environment be augmented with radiofrequency identification (RFID) tags to more easily detect location [4], but thesemust be available across the whole area to be useful. Another technology, not yetapplied to visually impaired navigation, is Skyhook Technology’s Loki system(www.skyhookwireless.com). This uses an updated map of available Wi-Fi accesspoints to determine the location of the user. However accurate location detectionis currently only available in a limited number of locations. There is not currentlya single technology that can always be relied upon to deliver accurate positioning.
Whilst it may be possible to use different technology to determine the user’slocation, these different technologies produce that information with differing de-grees of fidelity. When comparing the position of multiple technologies there maybe conflicts between their reported positions, or in some cases, no position atall. Whilst we will seek to improve the accuracy of detecting the user’s locationby using different, and perhaps multiple technologies, we also need to determinewhat should happen when an accurate position cannot be determined. The vari-able coverage of location based technologies is currently being studied in visualscenarios as seamful design [5]. Here the breaks in technologies, the seams, ratherthan being hidden from the user, are made more obvious, allowing awareness ofproblems before they occur. For example, our navigation system may monitorareas of poor signal strength and notify the user when it detects that it is at theboundary of that area. Further investigation into the exposition of seams in suchnon-visual applications may help to overcome many of the problems of currentlytracking the position of the user.
Mobile navigation presents a significant problem for visually impaired people.
From our requirements capture we can identify several issues that could be over-come with the constructive use of technology. However, from our initial proto-type system, there are several problems in the use of tracking technology thatcan easily cause the user’s location to be mistakenly detected. Whilst we believethat the position tracking of our system can be improved, there remains no onetechnique that can guarantee correct location, and any system will, at times,lose the position of the user. We propose that further research, on how best tomake the user aware of failings in tracking technology in non-visual scenarioswould significantly improve the quality of mobile navigation solutions.
1. Petrie, H., Johnson, V., Furner, S., Strothotte, T.: Design lifecycles and wearable computers for users with disabilities. In: Proceedings of the first international work-shop of Human Computer Interaction with mobile devices. Volume 1., Glasgow, UK,GIST (1998) 2. Loomis, J.M., Klatzky, R.L., Colledge, R.G.: Navigating without vision: Basic and applied research. Optometry & Vision Science 78(5) (2001) 282–289 3. Strothotte, T., Petrie, H., Johnson, V., Reichert, L.: Mobic: User needs and prelim- inary design for mobility aid for blind and elderly travellers. In Placencia Porrero,I., Puig de la Bellacasa, R., eds.: The European Context for Assistive Technology.
IOS Press (1995) 384–352 4. Coroama, V.: Experiences from the design of a ubiquitous computing system for the blind. In: CHI 2006. Volume 2., Montreal, Canada, ACM Press (2006) 664–669 5. Chalmers, M., Dieberger, A., H¨o¨ok, K., Rudstr¨om, ˙ design. Cognitive Studies 11(3) (2004) 1–11

Source: http://davidmcgookin.files.wordpress.com/2013/03/final-submitted-version.pdf