Pan-tilt sphere movement calibration

Hello, I working on a project that is meant to spray paint on a canvas. Here is a prototype of the robot working link.
The problem I am facing and want to over come is to create a calibration system and let the machine learn about its own movement and limitation by spray and see how far it can get.

The robot movement are based on a sphere due to be a pan/tilt, and the usual scenario is to place a canvas in front of it. The outcome should be that based on the calibration done the machine is able to measure the canvas, resize the picture, and spray. If the canvas is too big or too small a preview of the outcome should be thrown. Asking if I should continue.
I have been thinking to use a depth camera.

What I want to do is to create a model place various canvases in from of it and let the machine operate some sort of algorithm that would allow to it to train. So once the machine has been trained it can execute draws and learn upon them.

Any idea on who I can proceed with this project? I have a depth camera (xbox Kinect) that might be useful.

I am planning to use a Jetson nano with two cameras one detect actions, extract skeleton joints and the second camera is pointed to the canvas and execute the draw. I am wondering if a nano can do all of it. I do have another hardware which is the Jetson AGX so I thought to split the work between the two machine.


Do you need a runtime calibration or this can be done manually?
If manually is an option, maybe some checkerboard calibration can help.


Kinetic might be over killing, thought to use a camera, identifying the canvas and train a model. Is there and example where I can start quick?


Unfortunately, we don’t have a similar use case like yours.
Below are our Jetson samples for your reference:


1 Like

This topic was automatically closed 14 days after the last reply. New replies are no longer allowed.