The MaxConv problem

Hacks
1

I bought a small mill some time ago.
So far I’ve used it very little.
Part of the problem is software. The software I’m used to is made for 3d-printing and is not well suited for milling. The control board I have for the mill is also not very good for manual controls which makes it hard to just use it the old-fashioned way.

Anyways, I’ve been thinking of writing some hacks that generate g-code myself, and one of the interesting problems I’ve encounted in doing so is called the MaxConv problem. The basic thought is to have a profile that you want to achive and a profile of your endmill. Now “drag” these profiles against each other and calculate the toolpath.

This turns out to be a difficult problem to solve. Apparenly it’s NP-hard? Anyways, it seems like an interesting little problem to write some code for, so I’m probably going to do just that, benchmark it and see how it goes. :slight_smile:

Updates will be posted here.

1 Like
2

I love using my lathe and mill but am yet to use any automated system with them. What can I say I like the old do it yourself way😁 but I have seen some small mills that would do it for you and obviously saving time in doing it.
Seeing what you have created in the past I’m sure this will not be a problem for you for long👍.
I look forward to seeing what you get up to with this.

3

Did some poking around yesterday.
The problem isn’t NP-hard, not sure where I got that idea.
It is however N^2 in one dimension and N^4 in two dimensions.
It seems to me that it should be possible to compute the MaxConv at a smaller resolution and then use that as an approximation to figure out which areas are not needed to be calculated at full resolution, but I have not actually written that code yet.