Thursday, 29 August 2013

Updated the Matlab code to do localization in increments. Now moving on to send push notifications containing information regarding the nearest landmark- it's comments. Using the Google Cloud Messaging Service to do this.

Wednesday, 28 August 2013

Updated the phone app to log the sensor data every few minutes to the server. Now, moving on to the Matlab code to process these increments of data.

Monday, 26 August 2013

Updated the phone application to maintain a session with a server. Login data is stored on SQLite server on the phone. Had to change the database from postgres to mysql on the Server as the PHP readily available from apt-get does not support postgres connect functions. Now, moving on to implement incremental localization on the server. This can be done by storing the intermediate results in a matlab object and retrieving that as and when updates in location are received from the phone.

Wednesday, 21 August 2013

Updated the repo to include the Java Visualization application [This runs on the server and helps the mall owner to look into the data generated by each of the traces. Improvements to this will be done later]. Now, making the phone app session based, in order to implement incremental localization.

Monday, 19 August 2013

Update: Built a primitive Java application to help visualize the traces on the server, generate heat map and annotate the map with comments... Will update it with more interactive elements later, currently only images of the heat map and numbered comments are plotted on a graph. Now...moving on to implement the more important  incremental localization. This will enable us to update the location periodically, which inturn help us push data to the phone.

Sorry for the lack of updates this last week... Caught a viral infection while relocating and couldn't work much for this past week.

Friday, 9 August 2013

Localization update: Apply pedometer algorithm based on the activity recognized. There is visible change in location traced. For example when the mobile user stands and turns, the minor peaks created in the acceleration dimension was resulting a large change in coordinates. These false steps have now been identified due to its low variance in acceleration.

Now started working on a Java desktop application with Matlab component as its backend to visualize the data and analytics generated.

Sunday, 4 August 2013

Updated the code to rotate the location trace, thereby changing the orientation as and when landmarks are encountered. This way the coordinates of the the mall map will converge.

Thursday, 1 August 2013

Implemented testing of dummy landmarks, now moving on to changing orientation along with position at landmarks.