Show HN: BrachioGraph, an easy-to-build pen-plotter, driven by Python

Full details of the project, including all the code and documentation, are posted at https://www.brachiograph.art.

This is my attempt to do something with the Raspberry Pis and cheap servo motors I've had hanging around that's both genuinely interesting and also as simple and cheap as possible. Robotics isn't my subject, but I wanted to make something.

The total cost of hardware (excluding the clothes-peg, pencil and things you will find in a kitchen drawer) is €14. You could build the whole thing in about an hour, it's a fun thing to do with a child, and they love the fact that it's made from household materials.

One thing I like about it is that it's all 'wrong' - you are not supposed to use servos or electronics like this, and it's not what a 'real' engineer would do - but it still works well enough to be fun.

The cost and simplicity are really important because this project has come with me to events like PyCon Africa (http://pycon.africa) and PyCon Namibia (https://na.pycon.org) where not everyone has easy access to labs and hack-spaces and electronic hobby shops, and many people have to learn and experiment on much smaller budgets.

There's some custom Python code: https://github.com/evildmp/BrachioGraph and contributions and issue reports are welcomed. And it's on Twitter: https://twitter.com/BrachioGraph.

If you build one yourself, please send me a picture!

I'm a big fan of ink, since it gives better contrast. Can the python software be adapted to not drag across freshly drawn lines? Are there pens that work better in that context? I've got some parts laying around and I just might build this. Plus, it will make my brother-in-law sad that he spent > $100 for some kind of plotter he uses on a whiteboard.

Previous discussion about the documentation for this project: https://news.ycombinator.com/item?id=21281525

What are the "standard" algorithms for vectorizing bitmap lines?

I'd had a little art project where I wanted to draw lines on the wall and then have them captured by a camera and vectorised .. preferrably "minimized" to be as few vectors as possible. I couldn't really design an algorithm that was satisfactory and I didn't wanna just hack it with some heuristics

I see they're doing it here with an edge detector - anyone have any insight how it's done? Is this a solved problem in computer science?

A link to the project page with HowTos won't hurt - https://www.brachiograph.art/

Cool project!

Feature-creep: I can't help wondering if there would be ways to improve the precision enough to make for instance PCB's without increasing the complexity/cost to much..

Fantastic. Thanks. This will be my next project.