Careful reading of Vinay Venkataraman's paper ("Attractor-Shape for Dynamical Analysis of Human Movement: Applications in Stroke Rehabilitation Action Recognition"), yields these insights:
- The research team showed that phase space reconstruction could be used to classify movements in strongly periodic systems.
- The reconstructed phase space was further evaluated using a histogram of shapes.
Also posted in http://tangoproject.posthaven.com
Met with Xin Wei today and I'm kicking off the next phase of The Tango Project with a bit more of a general tone, incorporating into Rhythm Analysis with a focus on ensemble entrainment.
The first step will be to set up an event that will generate a variety of data, including:
- floor sensors
- cameras from many angles
- microphones
- XOSC devices on dancers (if they're willing) -- all dimensions
- iterative video (or process the video later in Max)
- other?
The rules of the first event:
- Three people (perhaps 2 dancers and one person playing music or beating drum).
- Do not set it up to have a purpose (i.e., don't tell people to try to synchronize, for example).
- Be careful not to imbue the event with my own expectations.
- Make it fun.
- Take data on everything.
Brainstorm with people prior to the event: Lauren, Jessica, Garth, Todd, Chris, the DS folks, Mohamed, Courtney, Garrett, Julian, Pavan.
Literature: Go back to the BWO, Manning, etc. to take philosophical underpinnings to the next step.
Develop research questions.
Schedule it.
Work with Mohamed and others to analyze data. For my own edification, how does the iterative video inform the transitions between self-organizing behavior and chaos, entrainment and intention? Can I detect in broad strokes the differences in state of mind of the dancers? How does this relate to dynamical systems analysis, if at all?
The primary activity this week was the first AMESA Brown Bag with the topic "Dancing with Dynamical Systems." There was a good attendance on the iStage for the presentation, including folks from engineering with specialization in dynamical systems, along with Synthesis staff and numerous AME grad students. Thanks to Garrett for getting the word out and coordinating this!
- I received extremely helpful feedback about the proposed recursive video experimentation:
- It might be useful to look at analog video because that may help to force stabilization in some instances.
- Look at audio because the data rate is much faster than video.
- Concerns that a focus on iterative video was actually making the problem of finding stability more complex and was not a traditional scientific methodology seeking to parse complex problems into more controlled chunks so that analytical methods may be applied.
- Human perception is not a reliable indicator of scientific truth so I should be cautious in depending on the human experience of video as an indicator of stability in dynamical systems.
Goals for Week of September 19, 2016
- Talk with Qiao about help with configuring MatLab for dynamical systems.
- Configure MatLab and run Vinay's program. Document results.
- Play with video feedback and running water. Make this fun and share results.
Thoughts and Observations
- My ongoing frustration with software is primarily configuration issues.
- It was extremely useful to do the Brown Bag since it forced me to review all of my notes. Still a lot to learn!
- I can walk a dynamical line between scientific experimentation and experiential media.
- Resist the temptation to judge dynamical systems. Stability is not always good. Strong feeling that the conditions immediately preceding bifurcations (and bifurcations) are of the most interest.
- I perceive that one of the issues in finding stability information in very complex time-series nonlinear systems is narrowing the search area. Perhaps the video (or other) feedback approach can be useful in giving broad hints and narrowing the search.
- It is extremely helpful to put ideas out in the AME community. The feedback helps me avoid becoming blinded by my own thoughts. :-)