Hi guys,
I’ve got a Universal Robot 3 collada model with its correct kinematics.

I need to edit the model but any modelling program I use (Maya, blender, unity, clara, sketchup and some terrible 3rd party program) won’t add the kinematics back in after exporting the model.

I’m fairly new to modelling especially any collada stuff (god damn texture mapping is a bitch) so any replies can be in an ELI5 format lol.

Thanks, Jared.

All I know about kinematics is that it’s the most underdeveloped and underutilized part of COLLADA. If anything implements it, I would expect OpenCOLLADA via its Maya and 3DSMax projects does… I would regard it as experimental and be surprised if much software implement the kinematics (IK) part of COLLADA. The 1.5.0 version of COLLADA was developed (I think) when COLLADA was already losing ground. Almost all implementations of COLLADA implement 1.4.1 and some will read 1.5.0 documents and make slight adjustments to support 1.4.1 features. But the things that are new in 1.5.0 never got put to use… that I’ve ever seen. (It’s hard to implement the features without documents. I don’t know if the Conformance Test Suite has documents for many of these features. Maybe it does for IK. But I would bet not… not without looking.)

EDITED: For the record, StandardDataSets\1_5\collada has kinematics folders (not physics) in the CTS (Conformance Test Suite). I might try to add IK to the COLLADA-DOM reference implementation… if they are good documents. I don’t know.

Hi Mike, Thanks for the reply
Yeah it does seem that way, Maya’s openCollada exporter is only 1.4.1 so doesn’t have any of the parts I need. I’ll have to look for that CTS to see if there’s any info about changing the models geom and updating the kinematics, need a proper scrub through.
Will have to try 3DSMAX but if it uses the same exporter then it might just be wasted time. So far I’ve been able to transform the geometry (so base is facing the camera) but it doesn’t let me update the base’s starting kinematics position.
I’m using ThreeJS to render the Collada by the way.

That’s interesting. I don’t know much about OpenCOLLADA, but the original authors of COLLADA were still active there, even last year I think. OpenCOLLADA is not like COLLADA, which is something I find very counterintuitive myself, since COLLADA is a whole lot to learn/implement, so it seems odd to couch it differently then itself, especially since you have to learn more about OpenCOLLADA. I’m developing new COLLADA tools for C++ but only at a glacial pace (see my signature) but I don’t think my work on it is proceeding at a necessarily bad pace.

I’m overhauling the original DOM, so what it is really is an XML Schema based heterogeneous container system, so that you can work with COLLADA exactly like how it is in a document, with really good C++ compile-time features and type-safety. I feel like COLLADA is so complicated that I wouldn’t want to try to do anything with it otherwise. It’s like a giant labyrinth, that I can’t imagine approaching with existing XML software. Which is why, I think it didn’t fair well back in 2010. Compilers are only beginning to be half decent at compiling the schema implemented with C++ templates. If you know anything about C++. And the Visual Studio IDE is really bad at parsing C++ (IntelliSense) so that it doesn’t help a whole lot to take in the schema. It fails at auto-fill, documentation balloons, and so on. So I think 2025 might be the right time frame.