[tjschweizer]

This is more out of curiosity than anything, but I'm wondering what topics I would need to learn to make a basic CAM software. I am thinking I would need to learn computational geometry, planning algorithms, and inverse kinematic control. What am I missing? Any advice? Again, I'm not planning on actually writing a full CAM package, but it seems like it might be interesting to get a few books to browse through, who knows what will happen.

If this is in the wrong forum please let me know

[Arjand]

Take a look at shapeoko, they made their own cam software in adition ti a very basic cnc machine. As basic as a non miller needs. Learn from thrm

[tjschweizer]

I can't seem to find any CAM software made by them, they recommend MakerCAM on their wiki. Which CAM software are you referring to?

Edit: Scratch that, looks like the Shapeko creator updated an old CAM software called PartKAM. So you are right, MakerCAM is Edward Ford's creation. Interesting.

[ger21]

It really depends on what you want your software to do.

Simple CAM programs just convert 2D CAD drawings to g-code. I wrote an AutoCAD macro that does this. Basically, all that you do is parse the file, and write the appropriate g-code commands. All that's needed for this is basic math, and maybe a little trig.

If you want the CAM to automatically calculate toolpaths, then you need to program algorithms to do what you want. These can range from very simple, to very complex. It's really up to you.

You shouldn't need to know anything about IK unless you're creating toolpaths for a 5 axis machine.

[Anniepoo]

I'm a professional software engineer who'se gotten old and bored with it, I'm having more fun making stuff that weighs something. I've written CAD and CAM software, I feel like I'm qualified to answer this.

You'd need to learn low level programming to do things like the motor drivers.
You'd need C and C++ cause that's what a lot of this stuff is writtten in, or in a language atop them.
You'd need to know computer graphics, for obvious reasons to display stuff, and because many of the algorithms you'd use would come from there
It'd be helpful to know some computer vision because many algorithms come from that field
You should know a modern functional language to actually write the thing in
You should know constraint programming because lots of algorithms like snaps and thinning depend on it
And you should know some logic programming because you'll end up needing to reason to resolve constraints
And if your CAM program ended up having an embedded language you'd want to know some language theory
You'd want some experience with numerical computing and numerical stability, you could get that by either taking a course in it or by studying some real time physics.

Now, the list above is sort of 'if I wanted to start my own auto plant what do I need?', You could certainly get by with a lot less. But yes, computational geometry (covered under a couple of my topics) and planning are topics you'll need. You'll only need IK if you plan on doing IK stuff.

[mountaindew]

I'm a professional software engineer who'se gotten old and bored with it, I'm having more fun making stuff that weighs something. I've written CAD and CAM software, I feel like I'm qualified to answer this.

You'd need to learn low level programming to do things like the motor drivers.
You'd need C and C++ cause that's what a lot of this stuff is writtten in, or in a language atop them.
You'd need to know computer graphics, for obvious reasons to display stuff, and because many of the algorithms you'd use would come from there
It'd be helpful to know some computer vision because many algorithms come from that field
You should know a modern functional language to actually write the thing in
You should know constraint programming because lots of algorithms like snaps and thinning depend on it
And you should know some logic programming because you'll end up needing to reason to resolve constraints
And if your CAM program ended up having an embedded language you'd want to know some language theory
You'd want some experience with numerical computing and numerical stability, you could get that by either taking a course in it or by studying some real time physics.

Now, the list above is sort of 'if I wanted to start my own auto plant what do I need?', You could certainly get by with a lot less. But yes, computational geometry (covered under a couple of my topics) and planning are topics you'll need. You'll only need IK if you plan on doing IK stuff.

All of the above gives a person a great deal of insight when using other peoples programs also!

[Anniepoo]

Bizarrely, sometimes not!

I've been climbing the same learning curve learning all the CAM software that everybody else on this forum is climbing.
Sometimes it helps - when I 'lost' my XML file (doh, had the software selecting the wrong config file!) I was able to analyze all the backups by writing logic software that reasoned about the config files, but
it sometimes still takes me a while to figure things out. I'm sitting here with a board with a pretty chewed up surface, trying to figure out what's off in my V carve settings right now, and I've spent a couple hours
struggling with moving data from Illustrator to Aspire.

Still, have fun! Read all that stuff! it's interesting! Half the people on here are hobbyist machinists, there's room for hobbyist programmers - it's all about love of turning the regularity of the universe into beautiful things.
If you run into problems figuring it out, ping me, I'll try to help.

[Arjand]

Is not from Edward Ford but from Inventables, the main part supplier for Shapeoko. Here is a link Using Your CNC Mill Sucks a lot Less With Inventables’ New App | MAKE