If I were doing ASL in Python, I'd probably take the following approach: 1. Determine the points where hand touches (or nearly touches) head or body. There are only a small number of standardized places, and a list is most likely available. 2. When the script is started, I'd try to locate those spots (as a vertex number) on the current figure, to the extent this is possible; then place little guide-props at those vertices. I'd also determine a "close but not too close" bounding box for each hand. 3. I'd then use the Point-At method to pull the appropriate hand[s] to those locations when needed, using the hand's bounding box against the guide-prop to avoid collision. This is, of course, useless for your method, but since I continued thinking about the subject, figured I might as well record my ramblings.

My python page
My ShareCG freebies