Chaos in the Double Pendulum

We built a double pendulum using rollerblade wheels and parts available at any hardware store. We followed, more or less, these instructions by Tom DeFayette and Gene Clark from Trinity University.  Here's a video of it in action:

Our goal was to use Matlab's image processing toolbox to extract a time series of the positions of the two pendula as a function of time.  By comparing two such time series, we would then compute Lyapunov exponents, as in the teaching paper of Shinbrot, Grebogi, Wisdom, and Yorke, available here.

Unfortunately in these lighting conditions, the exposures on each frame of our movie were too long to get nice crisp shots (the lower pendulum is smeared over a wide angle in many shots).

We were able to get limited results by attaching axis of the top pendulum to a rotary motion sensor and using LabView to capture a time series.  With the new high-speed camera in the lab, it should be quite easy to obtain an excellent time series.