Robot artist. How to translate a bitmap into a feature set?

Question

Do it yourself

I decided to build a homemade robo-manipulator capable of reproducing any text or image on paper in the most human-like way. If everything is relatively clear with the text, a big question arises with image processing. I imagined the implementation as follows: The image is converted into a set of contours using the Sobel method (1), the resulting black and white image is analyzed, and at the output we get a set of functions describing this contour (2). After that, these functions are transferred to the inverse kinematics algorithm, which, based on them, displays a set of positions for the hand (3). Can anyone give me a hint, dumbass. algorithm for constructing functions based on pixels (point 2)?

Answer 1

Search for "image tracing" keywords. Inkscape, for example, uses the open source program PoTrace.

Answer 2

opencv.org Have a
look, maybe you will find something useful for yourself.

Answer 3

Everything is very simple.
To control manipulators, the easiest (and most importantly more correct) way is to use g-code.
Read about how 3D printers and CNC machines are controlled.
And most importantly, there are plenty of ready-made hardware for them, for example, the most budgetary version of GRBL
Then the task is divided into 2
1) Convert the graphics to g-coge. To do this, there are many programs like ArtCAM
2) Posted g-code to control the stepper motors of the manipulator (GRBL)
Good luck