Home |
|
Project Ideas: Software
Here are just a few ideas for independent projects that would help advance the Wockets project forward (these would be appropriate for the Google Summer of Code, if the Wockets project is selected). If you have an idea for another project that you want to work on that is related to using mobile phones to detect physical activity in real-time, let us know!
For Google Summer of Code Applicants, here is the application format we suggest: Application.rtf
Your Idea: Useful Code for Using Accelerometer/GPS for Detection of Physical Activity
Who's working on it: Nobody yet
We have ideas, but we want to hear creative ideas from you about how we could create Wockets code that will use the phone accelerometer, GPS, or other sensors to detect physical activity type, intensity, duration, and location as best as possible ... for as many everyday or exercise activities as possible. What code can we write that will enable robust, real time, 24/7 detection of physical activities on phones (Android, Windows Mobile, Symbian, or other)?
Required: Demonstrable experience creating a sophisticated app on the mobile development target platform
Ideas? Contact us!
Porting Code
Who's working on it: Nobody yet
We have code running on Windows Mobile for real-time activity detection using the phone's accelerometer. We'd like to port this code so that it works on Android and Symbian. best as possible ... for as many everyday or exercise activities as possible. What code can we write that will enable robust, 24/7 detection of physical activities on phones (Android, Windows Mobile, Symbian, or other)?
Required: Demonstrable experience creating a sophisticated app on the mobile development target platform. Desirable: Familiarity with C#/Windows Mobile and pattern recognition and basic signal processing.
Ideas? Contact us!
Optimization of Feature Detection Functions used for Real-Time Activity Recognition on Mobile Phones
Who's working on it: Nobody yet
Are you into code optimization. We want our activity recognition algorithms to run blazingly fast on mobile phones ... especially Android and Windows Mobile. We need someone to scrutinize every line of code and help make our real-time activity recognition algorithms run as fast as possible and with rock-solid robustness. In particular, the methods that compute features from the raw accelerometer data (means, standard deviations, peaks, Fourier parameters, etc.) could use some reworking.
Required: Demonstrable experience creating a sophisticated app on the mobile development target platform; some comfort with C/C++ coding; prior experience optimizing code on mobile platforms. Desirable: Familiarity with C#/Windows Mobile and pattern recognition and basic signal processing.
More details? Contact us!
Fixed Point Implementation of the C4.5 Decision Tree Classification Algorithm
Who's working on it: Nobody yet
We have found that decision tree algorithms, in particular the C4.5 algorithm, work very well for mobile activity recognition from wireless accelerometer data. This project involves writing a highly optimized, fixed point version of the C4.5 classification algorithm that can be run efficiently on two mobile phone operating systems: Android and Windows Mobile. In addition to the straightforward implementation of the C4.5, we have ideas for ways we'd like to modify the algorithm to best achieve the Wockets goals.
Required: Demonstrable experience creating a sophisticated app on the mobile development target platform; strong interest in pattern classification algorithms. Desirable: Familiarity with C#/Windows Mobile and pattern recognition and basic signal processing.
More details? Contact us!
Building a Shapeful GIS Search
Database with Public Data
Who's working on it: Nobody yet
We want the Wockets system to use position data from mobile phones to be able to determine information about the user's environment that might explain or influence physical activity. Unfortunately, most GIS databases that store points of interest store them as zero-dimensional points. Researchers often need a 2-dimensional representation of these places so that they can decide whether or not a GPS coordinate pair is inside or outside of a region, such as a public park. Data about the shape of public spaces is available in public databases published by the U.S. Census, but there appears to be no easy way to search those databases and gather simple information from mobile devices (e.g., if I'm at this longitude/latitude, am I inside of a park?). This project involves developing server-side software that can interpret free, public shape data and allow coordinate-based search of shapes. An alternative strategy would be to figure out ways to make use of APIs from the free mapping services, but unfortunately most of them do not provide free access to some of the information that is of great interest to public health researchers.
Required: Demonstrable experience creating a mobile app on the target platform using web services. Desirable: Familiarity with GIS; familiarity with GPS/location APIs.
More details? Contact us! |