This was a contract gig for RouteMe2 Technologies, a Silicon Valley company that specializes in developing location based service (LBS) applications for Google's Android smart phone.

My job was to digitize shuttle route maps and timetables and create sets of files conforming to the Google Transit Feed Specification. There were 95 routes in total, mostly free shuttles operating in the San Francisco, and the work was done on a milestone basis over several months.

All of the bus stops for each route needed to be Geocoded and the timetable and fare information extracted.

For the former it was a matter of reading the stops from the route maps and trying to pinpoint them on google maps. Street view came in very handy for this and many of the bus stops I could actually see, which was great for accuracy. That part of the job was a little bit like a cross between aerial reconnaissance and "Where's Waldo?". I was even able to call some of the agencies to dig for more information.

For the time-tabling, KPDF has a great feature that lets you draw a rectangle around a section of text in a PDF and extract it as plain text. which ended up being a big time saver. I wrote a few short perl scripts to take a timetable in CSV format and covert it into GTFS format. I used an OpenOffice.org spreadsheet to keep the datasets straight.

The final destination for all of this data is an application called "PedNav", developed by RouteMe2 technologies. It's a google Android application that works a little bit like a personal assistant. You tell it where you need to be for the day and at what time and it will create an itinerary for you with instructions of where to walk, and which bus to take. The application won the first round of the Google Android Developer Challenge.

Url: http://www.routeme2.com/

Tags: perl, automation, applications, parser