You can spend a lot of time trying to think of a clever title for a post about a CNC chainsaw. But you’ll finally realize, what else can you say but “CNC Chainsaw?” [Stuff Made Here] actually built such a beast, and you can watch it go in the video embedded below. A custom chainsaw on a Tormach robotic arm. So it is more like a robot using a chainsaw than a conventional CNC machine.
Instead of an XY motion, the machine uses what the video calls an “apple peeler” method and uses the Minkowski algorithm to adjust for the size of the chainsaw. The video is an odd juxtaposition of advanced topics like the Minkowski and basic things like G code.
The chainsaw needed to keep a low profile, so it actually started life as the business end of a pole saw with drive courtesy of a skateboard motor. However, the cutting speed didn’t meet expectations, so that caused a change to the algorithm.
Some of the sketches to illustrate comments are pretty good and give a good idea of the issues involved. Like many things, it seems easy until you try to actually do it. Sadly, his saw got overheated so he finally switched to cutting foam instead of logs. With the right tool, however, the log should work. Maybe that’s version two.
Before you get too excited, you should know the arm used says it costs “under $20,000” so this probably isn’t a weekend project with your birthday money. Still, the problem-solving that went into it is interesting and you could probably figure out a cheaper way to do an arm if you didn’t buy it off the shelf.
Overall, this is a great story and the final product was surprisingly good considering how coarse the saw is. We liked the math and problem-solving behind the killer chainsaw CNC.
We don’t know if a different DIY chainsaw might work better. Not the first killer chainsaw robot we’ve seen, by the way.
Awesome project… XYZ PQR 👍
Ehh, needs way better software already available instead of having to spend a week or more actually programming something not terrible but not exactly great either.
Thought that robotic CRC multiple axis cutting was better than that now.
Some DCS would save a lot of chain saw rebuilding. :) Programming a bear? Turn the torque down bring Jethrow in and teach the points from a model. Chainsaw anything? Might be easier to use a tool center point IJK, material surface vector PQR, Typical XYZ, ABC and A^1 rotary table. Yep infinitesimal cut possibilities but you can only cut the log so many times and actually remove material.
Frigging fantastic ! Did not understand most of it, but loved the outcome !
Shane made a dog, his wife made a cat.
Robot with a Chainsaw!? Has this guy never seen/read any science fiction!? Kind of interesting though, if he can a big enough piece cheap enough (next winter) he should try ice sculpting with the arm. I’m sure this thing will need a hobby after its kills all humans.
At least the robot says “Stay back, I will kill you” before starting work.
Those robot arms are just as lethal without anything on the end effector, the chainsaw just adds a little extra reach. The only way to be safe is to always stay completely out of the “kill zone” when the arm is live.
… doesn’t he specifically move away from using the apple-peeler method because it would be too slow…
Reminds me of the “RoboCut: Hot-wire Cutting” system…
https://www.youtube.com/watch?v=lLKI0HWV3dc
Al, did you watch the video? The guy said he started with an apple peeler like method but quickly transitioned over to a custom slicer based on his own heuristics because the apple peeler method was very slow and the chainsaw wouldn’t be able to handle the wear from so much continuous cutting.
That chainsaw couldn’t handle wear from a minute of cutting!
Nice project but the video does seem a bit all over the place in terms of its target audience.
This project really makes me think about how people fund their hacking. I think a in depth look at how different people fund their projects would make a good article. Like for example, does this guy have a huge income and can afford to spend $20k without batting a eye? Will they be eating rice and beans for the next 3 years? Is someone else paying for it? Tax ride off as its for a YouTube channel so technically a business expense? Technical content is good, but there must be some financial tricks and such that some hackers know that others could benefit from. Cost is so often the most prevalent constraint in any project (professional or hobby) and it would be interesting if Hackaday looked at this.
He says at the beginning of the video that the manufacturer offered to lend it to him for beta testing. I guess that’s the kind of perks you can expect when you have more than 2.5M subscribers, no hack there.
If this doesn’t scream KILL ALL HUMANS!, what does?
I love it!
Now do it with a Husqvarna 3120 XP with a 6 ft long blade.
And to hell with the kinematics and leverage- we all wanna see it, YOU KNOW if you’re reading this you wanna see it.
Here’s a Yasakawa with a katana doing tameshigiri (test cutting practice) on hundreds of tatami mats simulating human corpses traditionally against a kenjutsu master with a sword to tide you over till then
https://youtu.be/O3XyDLbaUmU
Great video!
Thanks!
” probably isn’t a weekend project with your birthday money”
it was christmas money
https://www.youtube.com/watch?v=4VlwUT4pIH8
Love the contrast between $50 000 robot and total crap test setup. Kinda remind me Tony Stark messing with particle accelerator in his garage and shimming it with books ^^
Kudos to the original post (Stuff Made Here). Yes, he doesn’t really know state of the art algorithms for optimized 3D topology, but you can really project yourself in this trial and errors approach. I admire his relentless perseverance. I would be proud of this foam “doggish” sculpture.
This is how you lose a hand.
He had me at;
“This is one of those ideas thats so bad it wraps around to good”.
That almost sounds logical…….
Almost like modulo 2 logical.
Avoid local minima – use simulated annealing.
Do it faster – stop using python FFS!
Yup, spend 20x as long writing it in C for a 10x speedup.
Some of you really miss the point… There is not target audience, he makes stupid shit we all want to see because he can. Hes not trying to show what’s possible, hes just showing what dumb shit a company decided they’d be ok funding.
A robot arm with a chainsaw? Scary?
Nope.
I’ve worked in a factory where robot arms sprayed paint. Until it dries it is deadly. First exposure, usually no problem. But it changes something inside you. Second exposure, even a little drop you don’t even notice. You’re dead.
The people who worked back in that part of the shop were paid a lot of extra money for hazard duty. The rest of us happily stayed far away.
Paint is a ranged weapon. You can step just outside the reach of a chainsaw and be fine.
This is a super hard problem because the CAM software does not understand the cutting geometries of a chainsaw. There are a lot of dynamics in play vs a rotating tool bit. You have to figure out how to define the non-cutting parts of the bar. You have to tell CAM how to do a slicing cut like a normal saw (you can cut up or down in line with the plane of the bar but not sideways), then you have to explain the scraping cuts that can be made with the tip of the bar, and you have to explain the rules of a plunge cut with the bar. That is a tall order for CAM software that is built with rotary cutters in mind. You might be able to get a couple of those by starting with a ball end mill model. Think of the bar tip as a big ball end mill (radius equal to the radius of the bar tip) with a very light depth of cut. The rest of the bar could be considered as the (very large) shank of the tool.
That ignores a lot of other chainsaw cuts. For example, you can plunge cut with the tip but cannot move sideways unless the plunge is no deeper than the chains cutter height. An experienced chainsaw operator just knows all of this instinctively but think of programming that knowledge gives me a big headache. This is a case where it would be better for a human to manually control the arm and make the bear (probably really hard in itself) and record the moves to duplicate the model.
Probably you could digitize the moves made by a human using an ordinary chainsaw to make a bear just from a video. Adding some fiducial markers to the saw and work piece base would make it easier, of course.
Next step, you gather together a lot of these digitized recordings and train an AI system how to manipulate a chainsaw to make specific shapes.
Finally, you figure out how to integrate this into a CAM system to make arbitrary shapes. Should just be a simple matter of software :-)
Bar oil…..just sayin’
Step back or i will kill you!😲🤣😂
$2000 for a bear…just under $20000 for a mechanical arm…..$???? Time, labor, headache….
I want one!
Oh yeah, besides bar oil plunge types of cuts shou have been used later after more of the outside wood was removed. That would reduce the friction extending the life of the saw and not starting a campfire in the lab. It wasnt what he pointed out on the chain…….why didnt he get one of those really cute 5″ cordless chainsaws? Perfect size for that arm and they do better with plunge cuts too, just dont forget the bar oil