Re-purpose Industrial Robotic Arms

We must find out where you can acquire these industrial robots pictured above. Sure, you expect car companies like BMW to have a few lying around, which they used to make into a Twitter message writing robot. But Bungie, a video game company, to have one as part of an advertisement for Reach?

The former is just a scratch on the surface, with some pictures, but a much more decent writeup will be provided after September 12th. The latter has a few videos, and you can watch it recreate a monument with light ‘live’. And while both are impressive uses of old tech, neither answered our first question, we gotta get us one of these.

[Thanks Matt and FurryFriend]

53 thoughts on “Re-purpose Industrial Robotic Arms

  1. Those would lots of fun to have, that is until that one bad line of code causes the arm to spin around and smack you or your wife/girlfriend in the face. It could, however, be good practice for when we have to rise against out robot masters sometime in the near future.

  2. Repeat after me: NEVER stand within operating range of a robotic arm.

    NEVER. Many of them move very fast, are very solidly bolted to the ground, and WILL hurt you in lots of not-fun ways. You know, like, impact. Or pinching. Or crushing. Or stabbing you with that whiteboard marker.

    Movement instructions don’t include “place marker 1 inch from the whiteboard unless there’s someone standing in the way”.

  3. I’m fortunate enough to work at a place where we have no less than four older Fanuc arms that used to work for GM building cars.
    Our engineering department is working on getting one up and running, but it’s been a slow process as I don’t think we were given any documentation on ’em.
    Sure as sugar GE Fanuc doesn’t play nice with folks for free.

  4. I work with Daihen, Kawasaki, ABB, and Fanuc robots. Each robot platform has it’s own programing language with their own faults. I would say that the Kawasaki is by far the easiest to control remotely as you can connect via ssh and they have an api that you can access. Most of the platforms run either a version of linux, or embedded Windows. Older fanucs run DOS. It can be pretty painful to program these beasts, I have a gray patch of hair from the Daihens and wish I never had to touch one again.

  5. I saw some medium-sized arms on eBay for <1000$, but without any kind of controller (I think it didn't even include servo controllers?), seems like these make half the price for the little ones…

  6. If you feel like getting a used industrial robot I suggest looking for one from ABB. Their robotics studio software is extremely easy to learn and their Rapid Code is almost just plain English and yet massively powerful. A couple of months ago I saw one with 200kg payload rating go for €8000. That compared to the starting price for a bad boy of that size being more like @50,000 – €100,000.

    On the other hand KUKA robots are more awesome and kicks a lot more ass. (Not at all biases because I will be a programming intern at KUKA for 9 weeks this fall.)

  7. I work for ABB – the company that makes this particular robot. If anyone is really interested and has a load of cash burning a hole in thier pocket, here is a link:;

    Otherwise, here are a few nice videos of the robots:

    And of course, these robots also had a minor role in Terminator 4 (they were the mass producing the terminators at the end).

  8. ARMATRON! Man that was the shiz-nit when I was a lad.

    Bad timing though, as by the time I could get one I had moved on to more interesting and capable stuff to lust after. :)

    I’m pretty sure my company is serious about getting our Fanuc arm up and running (also as a marketing tool, actually.) which is cool, but bolting the thing to the floor was extremely interesting, and setting up a safe zone around it is a priority. (think huge light curtain box)
    The “good stuff” has been limited to mounting it, removing all the old GM-specific hardware and cleaning the thing off.

  9. i wonder why they used one of them bigger arms, usually for holding an led, there are smaller and much cheaper arms at kuka. also being german rocks sometimes, these robots are at my college every year, with people hiring and stuff

  10. They are very easy to find. Why dont you guys actually try.

    Industrial surplus is the search term to start with. The problem is you wont find one for $29.95 even used and 30 years old they cost $9800 for a cheap one that is worn completely out.

    P.S. they are not called “robot arms” in industry, so actually learn about what you are talking about and then you finally discover the search terms to use.

    “many arms like the ABB model above are paint line units.

  11. Here’s the real problem. Not trying to be a Debbie Downer, but here are some things to watch out for when trying to acquire old equipment.

    Even if you are able to purchase (or even get for free) a used arm, controller and (highly unlikely) all the appropriate cabling — you still need the software to make the thing work. Further supposing you manage to get the controller with a working hard drive with operational software on it; you still don’t have the motion control development system…that was on a computer that’s either still at the arm’s former home, or in a separate dumpster somewhere else.

    Getting the development environment is the Achilles heel here. You almost need insider help for that, because you almost certainly can’t afford to buy it, even if the manufacturer still offers it for sale, and even if they would sell it to you. And the software is probably licensed and locked, so a bootleg copy, even if available, would need to be cracked.

    Pendant programmable machines may be a viable option, *if* the pendant was included with the arm (probably not, because one pendant can be used with multiple machines and they probably didn’t dispose of all of them at the same time, and kept the pendant, or the pendant went to whoever bought the first unit for a higher price.

  12. @ Peter

    what the art collective illutron did was to skip that process alltogether, and only use the driver circuits from the control cabinet. they were much easier to hack into, and the old software living in the controller was replaced by an array of arduino’s controlled by an openFrameworks library with visualisation of the robot etc.

    here’s the reduced controller:

    see more about the process here:

  13. We use a bunch of nichi robots in our manufacturing process. We purchesed the old machines for a couple grand and have spent about 30 grand each in programming and programming manual purchasing. Right now, they pick up a stack of wood from a conveyor belt and stack it in a nice pile all day long. They pay for themself in a single year in cost, quality of work and dependibility vs a minimum wage person.

  14. The other problem that one no one has mentioned, is power! I had the option to pick up some Fanuc robots when my old job closed shop. Could have gotten everything that [Peter] mentioned. Manuals, software, hardware, pendants, cables, everything. Even some contacts with the guys that coded the units in production to pick their brains. Essentially you’re buying the units at scrap rate when they’re untested at closeout/auction. The problems for me though was power and space. Usually these things require 440 3-phase at some pretty healthy amperage. Even if you can get the juice to run it, my other problem was space. The vertical reach on the ones I could have gotten were higher than the rafters in my garage. So yeah, even if you do get all of the goodies, power can be a problem. If you can get tiny ones (less than 3′ reach) usually you can get 220V single phase servo controllers and you can cobble together a replacement controller software system. For the big boys, it just didn’t seem worth it. :-(

  15. I work at one of the these robot manufacturers, and there are many people doing cool stuff with these. They are pretty expensive, even when very old. So the best bet is to rent one or try to get the manufacturer sponsor you.
    If you have a really cool project, this isnt unlikely, especially Kuka is known for lending robots for cool projects. For example look at this link:

  16. Good to read all your comments.

    Powering a used robot for the hobbyist isn’t as difficult as it sounds. I know one guy who is powering a 125kg payload Kuka from a single phase to three phase inverter in his garage without problem. You are quite right about the cost of used units though robots do tend to retain a good percentage of their value throughout their lifetime.

    In the photos it is apparent that there is no safety guarding, this is because it is in a controlled environment. Robots are very dangerous and the chances of an unexpected movement in this project are increased because each character position in the sequence is calculated with an offset against the work object of the whiteboard. An error in calculation could see the robot veer in any direction, or even into itself!

    On the day it will be properly secured with guarding and will be nowhere near any person!

    Early on in development we did have some comedy pen/whiteboard smashing incidents but as any robot programmer will tell you that is just par for the course!

    Keep watching for the complete technical write up to be published on my blog.


  17. I recently heard about an advertising company that purchased several former GM robot arms to use as camera mounts. They mounted the cameras where the welders were, and now they have cameras that can rotate, track, etc. with high precision.

  18. If someone manages to get ahold of one of these I would use EMC (linux open source) to run it.

    It’s mostly used for home built CNC machines but is completely configurable and has some non-cnc example configurations (one is to control a hexapod robot)

    I just used it for the basic stepper motor on a desktop mill setup, but I didn’t buy a kit, just threw stuff together from the local surplus dealer so I learned a fair bit about configuring it.

    If I had one of these arms I would probably not bother learning the proprietary interface and just run the servo amps direct from a PC running EMC (necessary interface hardware can be built easily enough, or bought reasonably).

  19. Hmm if your looking to pick up a few of these I distinctly remember seeing some a few weeks ago. I can’t remember though if I saw a pair of arms at “Haulted Systems” or at “Weirdstuff”. Both stores are located about 10 minutes apart in Mountain View, CA. I’m not living in the area anymore bu there were two of these arms on the first row in the backroom. They appeared in good physical condition but I wasn’t really inspecting them a the time.

  20. @barry99705

    Unless you’re Tony Stark, I don’t think that’ll be an issue.

    As for the topic… wouldn’t a cluster of processors( one for each dof) be able to handle this with ease?

  21. @Rollyn01: Interesting thought there.

    I wonder if it could be as simple as grabbing the lines from each joint and rolling one’s own?
    As long as you know the specs on the motors and encoders it would just be a matter of juggling them the right way…I think?

    Sorry, lack of experience in this field is probably blinding me to something I’m not thinking of, but still…meaty food for thought.

  22. @strider_mt2k

    Juggling might require a stack or two to keep track of the motors position and speed. Other than that, you should be able to just have a master processor to distribute the commands to the cluster and they would in turn act like a spinal cord to make sure the arm moves in the right way.Also, like a spinal cord, the cluster would do some reflexive corrections by sending each other data about the spatial position of each dof( possibly by interrupts).

  23. The automotive transplant I used to work for would literally throw these away during model changes, it made me sick (Denso, Kawasaki, etc.) A contractor I knew told me of how they were disposed, they were lifted hudreds of feet into the air with a crane, and dropped into a scrap pile, smashing into bits. Then they were turned into pig iron……. Anyway, my woodworking CNC lived a former life as a cartesian robot, 600x400x100 mm, in an electronics assembly plant in Springfield, MO. Paid $500 plus about $280 shipping. Panadac Panarobo (Panasonic), 120VAC, 1200 W. Programming is similar to BASIC, very rudimentary but gets the job done. Came with a manual, controller, and prog. pendant. It took a good six months of looking on ebay to find it, then another six months of tinkering to truly get it running…..

  24. We made something simular to this for a show robot. Basicly we programmed our own fonts. Then all you have to do, is tell the robot witch letter/subprogram to use. What you could do, was build the font in the ygly/fast way, of printing the letters on a piece of paper, teach the robot the motions and that would give you your font.
    The words could be transferred via serial og ethernet, by a preprocessor that would build the entire program.
    As i recall, we had a routine running on the robot, monitoring the filesystem, and IF the module for the writ program was pressent, it would load it, execute it and delete it. ( i dont recall if this was a concept, or the completed solution).
    We even had it following freehand from a wacom tablet.
    the hard part for me, would be getting the messager from twitter,as i have newer done this, othervise a trivial exercise. Well anything is easy once you have done it :-).
    Btw in all fairness, the credit for the robot i describe go to HC and Q (you know who you are).
    Those guys have a nag for finding simple solutions to complex problems.

Leave a Reply

Please be kind and respectful to help make the comments section excellent. (Comment Policy)

This site uses Akismet to reduce spam. Learn how your comment data is processed.