<< return to Pixycam.com

Pixy precision approach method


Hello! I’ve bought a Pixy months ago and I’m just starting to play with it.

My idea is to implement a Pixy in a project in which I need very high precision. To simplify my idea and make it more easier to understand, my idea is to mount Pixy on the edge of a SCARA robot and to use it to positionate this edge on a wall.
To do so, I’ve thought to stick a reference to the wall, for example a 4 cm for side red square, with a 0,5mm blue square in its center. With an algorithm and the SCARA robot (let’s suppose that the robot have infinite precision) I’ll move the edge trying to have the square as centered as possible, and approaching to it until it ‘fills’ let’s say the 80% of the horizontal vision. At that moment, I will tell the robot to do the same, but now with the 0,5mm side square, adjusting the parameters of the robot of course.
With this method, I will theoretically reach the maximum precision when I’ll have the square sticker physically touching the lens of the camera. The minimum displacement of the square that will be detected by Pixy would be the maximum precision I could get, isn’t it?

I know this is a very theoretically approach, and at the real life I’ll have to take in account that there is noise and many things that will extremely reduce this precision, but I also think that I can reach a very high precision with this method.

P.D. I don’t care about the speed of the robot or it’s precision. Here I’m only talking about the Pixy precision.


Hello Miguel,
I’m not sure I understand your question. A 0.5mm square is very small. I would think that it would need to be very close to Pixy in order to detect it.