Experiment with different vision parameters

[charlesmackenzie]

Purpose
Right now, our vision system relies on the very vague tribal knowledge of a soon-to-graduate team member. We need to spend time playing with different inputs to the localizer in order to gain a more crystalized understanding of what they do, and a better set of "reasonably default" values.

Project Scope

• Play with camera resolution and understand its effects on frame rate and recognition range.
• Play with camera aspect ratio to find the actual maximum aspect ratio of our various cameras.
• Establish a step-by-step procedure for camera calibration, documenting what seems to have the most effect on accuracy.
• Experiment with accuracy effects of using a calibration file generated on one camera on a different camera of the same model.
• Find a reasonably reliable way to accept single-tag measurements that doesn't feel immensely hacky (they tend to be correct at least half the time).
• Find (experimentally or by reading theory) the relationship between localization accuracy and tag angle + distance.
• [ ]