Only Calculate Estimated Poses When Multiple Tags are Present with PhotonVision?

[gcschmit]

Describe the question you have.
For the PhotonVision specific implementation, is it intentional that the vision example project only considers pose observations when multiple tags are present?

Describe the reason for your confusion.
PhotonVision can provide pose estimates based on a single tag, as well as multiple tags.

Is your question related to a problem? Please describe.
While pose estimates based on multiple tags are usually more accurate, depending on the layout of tags on the field, this implementation may not provide pose estimates as often as it otherwise could. This could potential lead to frustration for teams who may not realize that the implementation only provides estimates when multiple tags are present.

Additional context
Regardless of the answer to this question, when multiple tags are present, the ambiguity will always be reported as 0, which may negatively impact the standard deviation calculation. For multiple-tags, an alternative is to use reprojection error as a measure of quality.

[jwbonner]

This is just an oversight, since I didn't realize that PhotonVision's multitagResult is not provided at all for single tags. 6328's vision system is designed to always provide field-to-camera transforms to the robot code regardless of the number of tags, and I just forgot to check that assumption. The commit I just pushed adds support for single tags, though it's unfortunately a bit more complex to implement in robot code since PhotonVision only provides the raw camera-to-tag transform. Definitely better than the alternative though.

With regard to multitag results having an ambiguity of zero, that's expected since including multiple tags solves the ambiguity issue (which is specific to square targets). Ambiguity is only used for filtering out results with high ambiguity, so setting it to zero just disables that particular filter. The standard deviations are based on the number of tags and average distance, so they will be unaffected.