These are geodesic kaleidoscopes made out of motion capture markers. The idea here is that both a human and a machine trained to recognize these markers using computer vision would be equally disoriented by viewing them.
Here are three of them:
See the rest of the kaleidoscopes here! (There are a lot more)