Cheapest way for converting a KUKA KRC1 to a milling machine

[robotrobot]

Hi everybody,

I have a KUKA KRC125/2 Tj with KRC1 controller and I want to change it to a 2D or 3D milling machine.
I'm looking a cheap solution, probably a simple (may be free!) software that convert CNC G-codes to KRL. I see some general questions about robot CAD CAM in other topics of this great forum but I decide to write this question because my robot is old model and I'm not sure the solution works. Also I hope some one who has similar experience with similar robot help me.
Many thanks in advance for your help.

robotrobot

[wes_mcgee]

I am not familiar with that actual robot but we use a kr100HA as a milling machine. The cheapest way would be to use kukaSIM pro and officelite to simulate code, and "script" the code from your CAD software(no CAM required). Notice I said cheapest, not easiest. This took us about 6 months to get working well, with the help of some very good CAD programmers(we use rhino3d). For the easiest, I would look into using masterCAM+robotmaster or robot works or KukaCAMROB(which requires a separate cam package I think). Those will cost more of course. Good luck, I think its a very achieveable goal, one you understand the complexity of KRL versus Gcode(the extra degrees of freedom are the problem, but they are also the solution!)

[robotrobot]

Thank you. Actually I want to deliver this robot to a customer as a CAD CAM system and I look to find  the cheapest user friendly software that help to make this system with a KUKA KRC125/2 Tj KRC1.
Is there any opinion or suggestions?

[SkyeFire]

Whew.  The problem is, I don't think there is a cheap and/or easy way to do what you want to do.  For one thing, converting from CAD/CAM to robot usually requires converting G-code CNC programs into robot code.  KUKA sells a software package (CAMRob) that does this, but it's not cheap, and your work is only as accurate as your initial robot setup -- to get machining-level work, you need a Really, REALLY good initial physical robot setup.  It'll probably require more than a few hours with a laser tracker and a skilled operator for same.

And watch our for accuracy.  An articulated robot turned into a CNC machine has a lot more reach and flexibility, for a lot less money, than any equivalent purpose-built CNC machine.  But the price you pay for that is less accuracy.  Articulated robots simply can't (yet) match purpose-built CNC machines for accuracy -- it's physically impossible.  There's a reason that CNC machines have so much heavy structural steel and are so much bigger, relative to their work volumes, than a robot in the same weight class.  There's been some work done in realtime guidance of KUKAbots using laser tracker systems or other similar devices, but those are a long way from "cheap", and they're probably not easy, either.

Still, if you don't need gunsmith-level accuracy, there's still a lot of CNC-type machining jobs out there that are within reach of an articulated robot... if you're willing to put in the "sweat equity."  First, if you want to offer this as a turnkey solution, you're going to need to decide on a 3D CAD/CAM software package.  Keep in mind that, in general, the cheaper they are, the less capable and/or user-friendly they are.  Then you're going to need to find a means of bridging the gap between the CAD/CAM model and the robot.  There are expensive off-the-shelf solutions, but you might need to do some serious research and roll your own.  You'll need to work this all up into a complete "plug&play" package... and then the customer will probably say something like:  "I don't like that CAD software, can't you use my favorite version?"  At which point you're back to developing again.

If handled properly, this is entirely workable.  But you need to set a well-defined spec and have a good idea of what you're doing.  One of the things I would suggest looking into would be the RepRap and Makerbot communities.  Their stuff won't be directly applicable to what you want to do, but that's where most of the open-source work in 3D CAD/CAM and CNC seems to be happening.  Digging into those projects will also help you get a handle on what kind of challenges you're taking on, and where you need to concentrate your efforts.  And the tools they're developing (FreeCAD, SkeinForge, etc) should be alterable to work with the kind of thing you're trying to do.

Good luck!

[najafabadi]

Hi Everybody,

Milling application with robots looks very interesting. I have a KUKA KRC1 to. I prefer to not generate a separate topic. Actually I think first step for building such application is generating a series of libraries for basic shapes. For example generating general KUKA robot program for a triangle path in a 3D dimensions and all other basic paths. Then we should separate each complicated surface to basic paths in next step.
Now my question is, does anybody know where I can find a free (or cheap!) basic shape path library for KUKA KRC1 (or other industrial robots)?

[wes_mcgee]

I agree with skyefire,I think it is going to be pretty critical what CAD software your customer wants to use. Robot code is just a series of points with orientations, and this data is pretty easily pulled from CAD software. We chose to develop it in rhino because there is a huge network of people in our field scripting rhino, but I think autoLISP could easily have been used(if anyone wanted to use autocad, but that would be terrible!), and I know of some people in switzerland using MAYA and python scripting to do the same thing.

I am not sure what the next post about "shape" library means. If you have a 2d contour in CAD, you already have all the info you need, you just need to extract the right values from the geometry and parse it into a generic SRC file. For use the CAM solution works well for milling, because it(robotmaster) has some nice features, but for contouring a 3d path, we can script it just as easily. I think having the KukaSIM/officelite software is critical there(and a very accurate workcell model), because CADCAM programs are not the kind of thing you want to have to simulate on the machine(if every program is unique). At least we are too lazy for that.