robot.has_any_barrel() is not robust

[touretzkyds]

If you run the attached program HasBarrelTest.py and give the robot a barrel, you'll see that robot.has_any_barrel() is very sensitive to lighting conditions. If the robot is facing a window, the barrel may be too bright to detect; holding your hand over the barrel to shade it will cause it to be detected. On the other hand, if the robot is in a shady area, the barrel may be too dim to detect; turning on your phone's flashlight and illuminating the barrel may cause it to be detected. In other conditions the barrel detection flickers on and off repeatedly due to the lighting.

The problem applies not just to has_any_barrel() but to the aivision system in general. Barrel detection is not robust. Unless the lighting conditions are exactly right, barrel detection is intermittent. This is causing problems for programs built on vex-aim-tools, which maintains a world map based on what the vision system is reporting. It's also a big problem for Calypso.

HasBarrelTest.py

[jpearman]

I think this is a duplicate of #16
I don't think there's much we can do about it.

[touretzkyds]

Okay. For #16 I was told the lighting was insufficient, so I had my office re-lamped. It's brighter now, but barrel detection is still unreliable. And there are cases where things are too bright. Is it not possible to adjust the color segmentation thresholds to accept a broader range of colors for "orange" and "blue"?

[jpearman]

Is it not possible to adjust the color segmentation thresholds to accept a broader range of colors for "orange" and "blue"?

Object detection isn't using any color thresholding (at least that we can control) but rather a model trained on several thousand images that's built into AIM. Perhaps send us some more images so we can see the environment, but lets do that on the Slack channel as this isn't really a websocket library issue.