.:: Previous Week ::.
Week 9:
~ Monday, July 23 ~
My window-matching program is becoming more and more complex. I really hope I will have enough time to optimize and clean-up the code. I tried plotting the source file vs. the extracted fragment. The result at a glance: synchronization is one of the major issues.
~ Tuesday, July 24 ~
To test whether or not the extraction is working properly I manually selected the data range and plotted the graphs. Things I was looking for: what part of the graph does the program extract the fragment from (i.e. does it match the “picking up the foot” fragment from the input file to “picking up the foot” from a different file or is the matching off?). It seemed that the program indeed found the fragment of motion that matched the correct acceleration pattern. After that I wrote the code, which extracted information for the specified data range.
~ Wednesday, July 25 ~
Meeting with Jessica:
- Takaaki presented direct mapping using two controllers
 |
| Jessica and Kate at a meeting |
- I also discussed the results of the previous mocap session: The previous mocap session was the first time we recorded motion using extra markers for the wiimotes. However, later it turned out that the results were problematic: the auto-labeling didn't work for the new markers and also a clapping motion (which we used as a synchronization step) caused most of the hand wiimote's markers to disappear.
As we later figured out, the occlusions of the markers occurred because they were too close to each other and the auto-labeling was confused because all the controllers were marked the same way. Thus, at the meeting we decided that if we change the marker locations for each wiimote and place them further apart from each other, we may get better results.
After the meeting Takaaki and I headed to the mocap lab. I thought it would be a good idea for him to see how the sessions are run and how to use my program's code to record the data. That way if he had any questions he could ask me directly instead of emailing me after I leave CMU. I let Takaaki get familiar with the program and run the mocap session himself, occasionally providing my guidance. I learned about the “release” version of the code and how to dynamically add directory to the filename.
After the mocap session I had an idea: if one “acts out” the leg motion (such a running) holding the wiimotes in their hands (instead of placing the wiimotes on the feet) and then try to match the produced acceleration pattern to the existing files, how would that compare to the actual leg motion?
~ Thursday, July 26 ~ * ~ Friday, July 27 ~
* Worked on my final paper: added the abstract, introduction and most of the background section.
* Consulted with Takaaki about the timing of the wiimote data recording: somehow the time-stamps were not reflecting real time but, perhaps, the duration of the loop?
* To my amazement I got a wonderful opportunity to meet with the creators of Alice www.alice.org and talk with them about their work's progress and future plans! My excitement was indescribable: almost like spending time with celebrities :-)
.:: Previous Week ::.