Vision reliant on proper gyroscope input, gryoscope can drift.

[spderman3333]

When testing at the Bloomhouse field, we noticed that the position estimates returned by Photonvision were, 1, very jittery, and inaccurate, 2, seemingly "inside" the field elements. The robot also seemed to think that it saw the Apriltags from their "backs".

The issue is related to hardware, or, more likely, there is an issue with switching robot origin to red from blue.

In-order to prevent future issues like this, we should consider using only a single origin (blue), as defined in the WPILib documentation.

[spderman3333]

Issue found, and reproduced. Solve with PNP requires proper gyroscopic input
Us crashing the robot into a field element lead to the gyroscope drifting, thus becoming inaccurate.

[spderman3333]

Next steps:
Test other estimation strategies, develop a way to detect fault rotation/drifted gryoscope.