Directory of Resources
Flocktracker is a smartphone based application designed to read your traditional pen and paper survey and convert it into a dynamic, interactive smartphone-based survey tool that combines dynamic trip data, along with anything from ridership to counts to vehicle speed to create rich, robust data sets that lend new insight and analytical potential to your research endeavors. This data is visible live and, combined with a forthcoming Dashboard system, will build off of current visualization provisions to enable operators to monitor the locations of their surveyors, as well as broadcast live, up-to-the-second updates on desired aggregate data metrics.
Site that is currently the home of our staging server, where the latest build of our online web app and dashboard system resides. Eventually this will be the home of the full application.
Project site from my graduate research that was used to hold the last version of the application as well as documentation including a user's manual that paired with this early iteration of a fully customizable version of Flocktracker. Last updated early in 2014.
Site that went along with research in early 2014, documenting the application's improvements in the field, in the south of Mexico City, over previous years' research and in field deploys.
Fairly well updated page where we connect with fans and users (mostly in Mexico) and keep followers updated on project advancements.
Various video associated with the project, primarily used for competition submissions - though all are fairly early stage content.
Origins of Flocktracker
Initial Development
Flocktracker started as a project under Albert Ching and Stephen Kennedy in the Planning Department at MIT. Albert had worked at Google prior to entering the Master's program at MIT and had extensive experience in India and Bangladesh. His travel and work there made it an area of focus that he set about expanding upon through his research with Prof. Chris Zegras. Through Albert's thesis work, he and Stephen scoped out and piloted an early version of Flocktracker using MIT's App Inventor software, which enables the creation of simple applications for Android using an intuitive graphical interface to abstract code development.
The Beginning of My Work with Flocktracker
It was at this point that I joined on, in the Fall of 2012. With Albert and Stephen, our first goal was to stabilize the alpha version of the application and use it in an aggressive first full roll out - the mapping of the bus network of Dhaka, Bangladesh. This project occurred over the January IAP period in 2013 and was funded in part by a successful Kickstarter campaign. On this project, I served as the lead coordinator in Dhaka, arranging the teams and managing the data produced by my team members. I also spent days riding the buses myself, assisting in the production of the map. Over the course of the Spring of 2013, Stephen and I parsed through the data and stitched together a complete bus map of Dhaka, which was printed thousands of times and distributed both physically and digitally throughout Dhaka in May of that year.
Mexico City
In the summer of 2013, Prof. Chris Zegras and I received funding through MISTI-Mexico (a department within MIT that supports international research) to work with Urban Travel Logistics (UTL), a transport engineering and planning firm located in the D.F. There, we designed an iteration of the early Flocktracker tool used in Dhaka to perform an analysis on the performance of CETRAMs (high capacity intermodal transit stations positioned around the periphery of Mexico City that acted as hub points between major transit route termini for BRT and subway lines, connecting riders to the collective transport vehicles that blanketed the farther reaches of the metropolitan region). During this project, we realized the flexibility and potentially powerful usability that could be achieved from Flocktracker, if it was made able to consume a generic survey JSON, thus enabling the application to perform on-the-fly, dynamic survey composition as deployed from some central control point.
Dynamic Survey Builder
Work on the current version of Flocktracker began in the fall of 2013. The goal was to produce an Android application capable of consuming a standardized JSON file, thus enabling it to auto-build surveys based on this input. This format is now called flockSON (pending some clever name improvement). Documentation of Flocktracker's flockSON design and capabilities is maintained on this ReadTheDocs resource page. By January of 2014, this version of Flocktracker was completed. This new version of Flocktracker was written in Java and built specifically for the Android (as opposed to utilizing the limited capacities of MIT's App Inventor, which resulted in an unfortunately unstable project).
Building out the Complete Tool
While the flockSON design was functional and the application had demonstrated its utility, it was severely limited by the technical requirements necessary to utilize the application. For one, creating surveys mean understanding JSON. While JSON is fairly intuitive, writing your surveys in it and making sure you have the correct parentheses and commas in the right areas is not, particularly for nontechnical users. A further limitation of the application is the use of Google Fusion Tables as the backend data base. While it works fine enough for performing one off surveys, it makes setting up projects overly complex, and also limits usability in locations like China. Additionally, the use of Fusion Tables makes Flocktracker dependent on a tool that is outside of our direct control. As a result, work began in the Fall of 2015 on creating a suite of tools to support the Flocktracker application. This includes a survey builder, that dynamically assembles the JSON through a user friendly interface that features drag and drop mechanics for assembling your survey. Work is also under way, presently, on completing an internal API that supports utilization of Amazon AWS to avoid reliance on Google's Fusion Tables service.
Flocktracker
