Week 1: (05/22/06 - 05/26/06)

Final exams are over and summer begins! After one week of relaxation, I finally got to pull up my average sleeping time. Since I am staying in CMU, I don't really need the "adaptation time" and I can start working right away. As this is the start of the research, I have set up my research goal and started to subdivide the whole problem into different parts. At this stage, I have to figure out the primitive actions of burger flipping, then the most appropriate data structure and algorithm to determine a correct burger-flipping sequence, and how to recognize the flipper's motions in order to separate them according to the primitives with just a small number of markers and 2 cameras.

After studying the project requirement and motion capture data, I have found that the burger flipping sequence would at least consist of the following primitive motions: placing the burgers onto the cooking plate, salting, flipping, pressing and picking them from the cooking plate to the serving plate. Then, in order to find out the most appropriate data structure for representing the whole sequence, I have tried drawing out finite state machines. However, because of a lot of restrictions on the burger flipping sequence, using finite state machines does not seem to work at the moment.

On the other hand, since I have to find out the way to recognize different motions in burger flipping, I have tried to extract and compare different information from the motion capture data. In order to do this, I tried to extract the 3D positional data of each of the 3 hand markers in the c3d files using c3dserver and MATLAB and plotted position trajectories for different motions such as:

 

Flipping

 

Salting