Touching distance threshold is a completely arbitrary number

[boyleconnor]

The touching distance threshold of 50 is a completely arbitrary number that is probably way too high (i.e. most objects would be considered touching even if they are fairly far apart).

I'd like to do base this number off something empirical, but I'm not sure specifically how I should go about finding a good threshold. We might have to wait until Blake sends us a new multi-object curriculum.

[boyleconnor]

@spigo900 @lichtefeld do you have any strong opinions on how we should determine this number? I suppose I could come up with an experiment to determine a good number from our existing dataset

[spigo900]

@boyleconnor Not particularly. It probably makes sense to try running on the small M4 curriculum. You could look at the results there and tune based on eyeballing if the objects are touching -- maybe annotating the situations somehow for touching.