The test sequence worked as expected, but since it is not a full 360 degrees, it is not particularly well fit to be a full panorama
The sequence taken with the Kaidan head did not turn out as nice as I had hoped. The reason for this is probably that there are objects in the close foreground and the far background that cause the Lucas-Kanade algorithm to incorrectly identify the motion of either foreground or background objects (since only one may be considered). Additionally, the reflection of the view on the window and doorway cause the Lucas-Kanade algorithm to line up the reflection instead of the window and doorway themselves. Lastly, there apparently were not enough images to create a full panorama (the first and last images did not overlap).
The sequence taken by hand actually turned out fairly well, although it is rather blurry since I did not pay a lot of attention to rotating the camera about its optical center and other such considerations that must be made in order to produce a perfect panorama. This panorama is also not a full panorama since the last image to finally overlap with the first image was not available since I accidentally did not take that picture.