Home alone with my vision machine learning robot friend. It thinks I’m sitting beside a banjo, acoustic guitar, fiddle or wine bottle. Runs off 2 raspberry pis. One controls the pan and tilt motor which houses the camera. The camera is attached to a movidius myriad 2 VPU for neural computing and another raspberry pi. I got the movidius hat from the Google AIY vision kit. I put the whole contraption on a battery pack so it would be portable.

I was planning to mount the whole thing on top of my old pi racer project so it would recognize things as it drove past them and have some self driving ability. The image recognition I am running is all done locally on the pi zero. I would like to think that is causing the accuracy of the image recognition to suffer. I guess my pimenator project is going to have to wait a little bit more.