Powered by Blogger

Saturday, April 07, 2007

More on Mainstream GPS Solutions

Continuing in my search for a mainstream GPS solution, I started looking into freeware, shareware and other very low cost solutions.  To begin with, programs selling for a few bucks or distributed freely likely cannot include a speech synthesizer as Microsoft SAPI support on Windows Mobile devices is incomplete and, at least through WM5 MS does not include a synthesizer.  

Thus, a $30 product, like VITO SmartMap from Vito Technologies has no value for a blind user as it has a purely visual display.  Interestingly, Vito SmartMap uses all vector graphics for its mapping solutions and sells maps of entire continents for $4 per continent.  I have no idea how complete the maps from Vito are but starting with a set of vector maps would make conversion to SVG far easier than with bitmaps which, in turn, can be rendered as a tactile map on a View Plus Tiger or other embosser.  To this end, I sent an email to Vito Technology to see if we might be able to gain access to their maps to use as the data for a tactile map project that someone might like doing.

The next stop on my search for an off-the-shelf solution brought me to a list of 17 Freeware GPS Solutions for Windows Mobile Smartphone.  The link I provide brings you directly to the Windows Mobile Smartphone GPS page.  This site, however, has literally thousands of programs in dozens of different categories for portable devices, including WM 2003, WM5, WM6 (although I don’t know of any WM6 devices for sale yet) and Symbian.  I will be revisiting this site to find more programs that work with Mobile Speak Smartphone but anyone who has a WM or Symbian device (including PAC Mate) with a screen reader installed should check out this site if you want to find all sorts of stuff that might work for you.

As I explored the site further, though, I learned that they actually have 3 GPS programs listed and about 14 or 15 other Smartphone applications that somehow got miscategorized into the GPS section.  I didn’t look at the process one goes through to list a product on this site so I don’t know why GPS came up with so many false positives.  The bulk of the programs that do not deal with global positioning systems do seem to have something to do with health and/or fitness so maybe there exists another meaning for GPS?

None of the GPS programs I found on this site seem suitable for use by a blind person.  

The first GPS program on the list, however, is called PerDiemCo Location and Tracking.  This program does not provide navigation or mapping instructions at all but, rather, lets you keep others informed of your exact location.  While I don’t think this has much value for a blind user (I didn’t test it for accessibility either), it may serve a good purpose for elders with dementia and others with cognitive impairments who may get lost or badly confused from time to time.  

Oddly, as I searched in the GPS category, I did find that some of the health and fitness programs might interest a blind user.  I haven’t tested any of these for accessibility but the list does include a few programs designed for diabetics which our readers should know has a large cross over with people with vision impairment.  I’d be interested in hearing if anyone has tried these programs with a screen reader on a mobile device and how well they work.

Thus, my search for a GPS solution that I can use on my WM5 Smartphone seems to have come up empty.  None of the open source GPS programs run on WM5 Smartphone devices so I can’t simply tweak something into working under 5 mph.  

I will now broaden my search to include standard Windows Mobile 5 devices like an iPAQ and, if I have time, I’ll try out some WM 2003 programs also on an iPAQ I have available to me.  

There is an open source Symbian GPS solution designed for blind users called Loadstone.  I might dust off my Nokia 6600 to give it a test drive.  I have heard anecdotally that the maps used by Loadstone come from the US Census Department and contain huge missing pieces of some areas.  There are a few online, “open source” map databases that volunteers fill in with data as they discover empty patches.  For people who live in places where these databases have pretty solid maps, they could provide a low cost mapping solution.  

If someone wanted to build a web server side GPS navigation solution for blind users, google maps provides a very solid API but, even for users who have unlimited data packages on their cell phones, google might be a difficult solution for making an application that runs on the handset as I haven’t found SOAP wrappers for google maps which would mean someone would need to perform the tedious task of generating such.

So, why do the otherwise accessible GPS solutions contain the 5 mph minimum speed for positional accuracy?

Virtually none of the GPS solutions I tested seem to pay attention to WAAS encoding.  WAAS provides much more accurate positional information than does traditional GPS.  WAAS, however, has a few major problems.  First, the WAAS system only exists over the United States, a bit of Southern Canada and Northern Mexico.  Thus, any GPS software vendor who wants to sell their products internationally cannot rely on this system.  Secondly, even in the US, a system cannot always count on the availability of WAAS signals as the US government can optionally turn the system off for national security purposes.  

WAAS, when it is available, consistently provides 1 meter accuracy.  This is good enough for a pedestrian who is paying attention to his whereabouts.  Standard NMEA GPS can, if enough satellites can be found by the receiver and there is little distortion in the satellite signals (tall buildings, mountains, canyons, thick cloud cover can all distort signals) also provide 1-2 meter accuracy but, unless the user is on water or in the middle of a field on a sunny day, such perfection rarely exists.

Most map data is also far less detailed than GPS, with or without WAAS, is accurate.  For instance, a map data source like Microsoft’s MapPoint (one of the very best) contains the GPS points of every intersection in its coverage areas but addresses in the middle of a block are calculated through interpolation.  So, when I wrote some test code for an open source GPS program that I may release that I’m currently calling “Freeway GPS” that used Microsoft’s MapPoint as its data source, I would get four addresses returned through the MapPoint SOAP interface when standing in my own front yard.  Using various heuristics, like calculating the distance from my exact GPS point to that of each of the addresses returned, my program could fairly well guess which address was closest to my actual location.  Unfortunately, because the addresses are calculated through interpolation, it was impossible to get my actual address and, often, the result would be a address that didn’t actually exist but, rather, stood between two real homes.  

For a map database to be truly accurate to every specific address, it would need to contain a bounding box for every separate address in its coverage area.  Thus, every spot in the map would require a polygon with at least four vertices into which one would need to calculate whether or not a GPS point falls.  This would be both processor intense and require an enormous amount of data on the server.  In turn, the extra processing and database hits would slow the system down to a point that anything moving even as fast as a bicycle might be well past the location reported.  As, to make any real money, GPS systems must work for motorists, rapid response is essential.

For these technical and economic reasons plus the lack of reliability in the WAAS system, truly pedestrian friendly GPS programs may still be left to the AT companies and the high priced AT market as leveraging the extra effort would be difficult to justify for a mainstream company.


I don’t understand the science of attention well enough to comment on the ideas that Will Pearson posted yesterday.  Even if I am distracted by a speech system, I still think I would like a GPS navigation program that I could easily carry with me.  I’ll pay extra hard attention not to walk into obstacles while listening to directions from the software.

J. J. posted a link to another GPS program for Windows Mobile 5.  As it does not have a Smartphone edition, I will take a look at it when I start to survey PDA solutions along with Mobile Speak Pocket.

-- End


Post a Comment

<< Home