Embedded Files
Our Algorithm is separated into a number of parts:
Figure 9
First, we have the "Perfect Toss". The way we acquired the data for this "Perfect Toss" was an averaging of different completed pancake flips. We took measurements for 4 similar, completed flips, and then averaging the acceleration and orientation values for all of these and creating one "perfect" toss. This data is then used to calculate the position of the pan during the pancake flip, which is what we actually plot.
The second part of our algorithm is loading and calculating the position of the toss we are comparing. This is done with the same exact process than Part 1, which gives us a matrix that has the position and the FFT of the toss we are testing.
Figure 10
Figure 11
The third part of our algorithm is to take the FFT's of the perfect toss and the testing toss and calculate what the difference between them is, and finally taking an average of all of these differences, across the two dimensions we care about. This is what determines the color of the plot in the following step.
You can see two examples of these FFT graphs on either side
Figure 12
Figure 14
Finally we plot, point by point, the position of both out perfect and testing tosses, with the color of the testing toss being defined by the value of the difference of the FFTs.
This lets us see the differences in the plots, with the coloring showing us plots that were closer to the perfect one or farther away. The closer to green the better it is, while the closer to dark blue the worse it is.
Figure 4
Figure 5
Page updated
Google Sites
Report abuse