News for hackers. Stuff that matters.
Re:We taught robots to carve marble sculptures by voidUpdate on Thursday November 19, 2026 @07:36AM (#50559305)
What part is being "taught"? It looks like a fairly standard CAM pipeline to me...
Re:We taught robots to carve marble sculptures by GrinningFool on Thursday November 19, 2026 @07:50AM (#50559442)
In the before times, we might say we "taught" a system to do something and mean that we designed/implemented the feature.
Re:We taught robots to carve marble sculptures by suddenlybananas on Thursday November 19, 2026 @09:44AM (#50560790)
Did we? "We taught a computer to send emails"? Sounds bizarre to me.
Re:We taught robots to carve marble sculptures by GrinningFool on Thursday November 19, 2026 @02:02PM (#50564617)
Heh. Well, some of us did, anyway.
Re:We taught robots to carve marble sculptures by bazzargh on Thursday November 19, 2026 @06:44PM (#50568214)
I'm not sure if it's the current style, but in the early days of git, commits to git itself and linux often had the subject line of "Teach X about Y". https://github.com/search?q=repo%3Agit%2Fgit+teach+about&typ... - I liked that, and used it elsewhere.
Re:We taught robots to carve marble sculptures by amelius on Thursday November 19, 2026 @08:32AM (#50559868)
Maybe they used an LLM?
Re:We taught robots to carve marble sculptures by Overdrivr on Thursday November 19, 2026 @09:10AM (#50560337)
Honestly for CAM you absolutely need deterministic output (Speaking from experience, I built automated CAD to CAM pipelines in the past using Python+Siemens NX). IMO the true innovation is the algorithm that they must have built that turns a target voxel into a 5+-axis machining toopath, but that's good old fashioned algorithms, and also getting to a 100% automated CAM generation. Achieving 95% automation is easy, 100% is very hard
Re:We taught robots to carve marble sculptures by voidUpdate on Thursday November 19, 2026 @09:29AM (#50560592)
There are a couple of mentions of "machine learning", in the parts about generating the toolpaths and connecting the toolpath segments into a continuous path, but I'm still not sure I'd call it teaching robots
Re:We taught robots to carve marble sculptures by ourmandave on Thursday November 19, 2026 @10:13AM (#50561171)
Are they shaping the stone to a figure, or removing the stone to reveal the figure in the block?
Re:We taught robots to carve marble sculptures by indoordin0saur on Thursday November 19, 2026 @10:56AM (#50561764)
Everyone knows that you'll get more hype and funding interest if you somehow connect your work to AI.
Re:We taught robots to carve marble sculptures by KaiserPro on Thursday November 19, 2026 @12:06PM (#50562803)
Yeah I was about to say the same thing. just looks like normal CAM to me.

Good efficient CAM is hard to do, and looking at how dense those toolpaths are, I don;t think its thats efficient. It also seems to only use one tool. I assume thats because anything bigger will cause the arm to deviate toomuch.

Re:We taught robots to carve marble sculptures by MisterTea on Thursday November 19, 2026 @12:34PM (#50563200)
> I assume thats because anything bigger will cause the arm to deviate toomuch.

I've been around a lot of CNC machines, wood routers, metal cutting, plasma/water-jet/laser cutters, laser welding, etc. The key to accurate motion is rigidity and heavy CNC machine tool frames that push a tool into work are hunks of cast iron or steel structures encased in cast epoxy granite. They ride on box ways or heavy linear bearings. They are MASSIVE.

Those robot arms have the rigidity of a wet paper straw by comparison. IMO, this should be a two step process where traditional machines with big tools hog out the rough shape then the robot arms carve the final details.