Today I wrote a script to query the translational information for the shoulder and the rotational information for the forequater segment. This was the segment where the elbow was the pivot point. I focused on determining how the angle of the elbow changed and if there was an obvious cycle for the movement.

Below is an image of the angle’s values over the video frames:

Now I’m looking at information on convolution, trying to dust off some EE skills. Here are some site that I need to finish going through (Google “fitting functions from data”):

- http://terpconnect.umd.edu/~toh/spectrum/CurveFitting.html
- http://reference.wolfram.com/mathematica/ref/TriangleWave.html
- http://www.physicsforums.com/showthread.php?t=82226
- http://newtonexcelbach.wordpress.com/2011/02/04/fitting-high-order-polynomials/
- http://stackoverflow.com/questions/1073606/is-there-a-one-line-function-that-generates-a-triangle-wave

The elbow’s angle looks similar to a saw-tooth signal. Go over convolution stuff.

ALSO, finish with Huihua’s expressions and determine if they work or not. Over the weekend (once the ray tracer’s done) finish getting the rotoscoped rotational data for him. Not sure if there’s any data we can use from him for the quadruped function.

======================================================================================

Consider speed caps for the different gait “states”. Elaborate in another post. State machines, speed of gaits, transitioning between states and the acceleration increase and then decrease after next state is reached. Also how to layer the movements of the different parts once they’re defined. LOTS MORE TO DO!!