Relative motion in Animation

[arnou-verfaillie]

Hello

Is it possible to make an animation movie that contains contact sets or relative motion? What I'm trying to do is:

I have a robot that handles doors, but I want to make an animation for further analysis, however in the animation you can only move components as a complete part. I want to move my frame of the robot so that the robot follows the motion.

I have recorded a screencast of what happens in the animation an what I want in modelling.

I've also tried the motion study, but it's very hard since if I give a direction to one joint all the others move in an unwanted way. I have attempted several times with motioin study, but it just bugs out.

Thanks in advance!

[arnou-verfaillie]

This is what I mean with a wonky motion study

Anyone an idea why this even happens?

[jeff_strater]

Hi @arnou-verfaillie,

Motion Study is the way to go.  Not only can you get the motion you want, but you can also render it in a high-quality cloud-rendered video.

As to the problems you see in Motion Study, I'd have to see the model to be sure, but my guess is that when you add this point to the "orange" joint in the Motion Study:

and the "blue" joint moves as well:

That is because there is a Motion Link between those two joints.  So, when you update one, the other must be updated as well.  Of course, there could be a bug in Motion Study, but, the Motion Link is the most likely culprit.

Jeff

[arnou-verfaillie]

Ok I will check this out, thanks for your reply!

[arnou-verfaillie]

Hi @jeff_strater

There are no motion links active in my complete assembly or in my original component "Robot". The only active motion link is between two different components. 

Here is the assembly: http://a360.co/2feua2O (if you could download or copy it and not save it? I'm still working in it).

EDIT: Also how is this even possible?:

In some other attempts to construct a motion study I've had this situation again, where a point is set, but even in the same step it goes to another point. If I recall correctly there should not be a singularity in this point. This is very confusing

[jeff_strater]

Hi @arnou-verfaillie,

Thanks for the model.  I have a copy now, so it won't affect your work.  Yes, you are right.  This looks like bad behavior to me.  I was able to reproduce it.  I think it's a bug (or at least undesirable behavior) in the Fusion joint solver.  You can see kind of the same behavior by just selecting one of the joints in "Robot", and doing "Drive Joints".  It appears to happen only on the first increment.  If I drag one of those joints, the other also seems to move, but just on that first increment.  After that, it is fine.

Screencast:

So, what I think is happening in Motion SNote that this effect is being multiplied, because that "first increment" in the Motion Study UI can be big.  Note that it does not happen if I put the Motion Study inside the "Robot" component itself, so it must be a situation relative to the top-level assembly only.

Let me dig deeper into this.  It is very strange, certainly.

Jeff

[arnou-verfaillie]

@jeff_strater

Yes it's very strange indeed. I haven't tried to actually do the motion in the Robot component by itself, that was a good idea. I hope you will be more succesfull on finding a solution, because I have tried many things to solve it: with/without contact sets, the order of which joint I choose,...

Thanks in advance for your help!

Arnou Verfaillie

[jeff_strater · Accepted solution]

Ah, I think I understand now.  It was right in front of my eyes.  The source of this behavior is the contact sets that you have enabled in both the top-level assembly.  This actually makes sense.  When you move one joint, and there is a contact set enabled, it can easily cause other joints to move, because the component must be moved so as to not violate the contact.

Personally, I don't use contact sets except in very simple cases.  The performance is usually very bad on any non-trivial model, and there are other problems, such as moving too fast can actually cause the contact to be violated.  Especially if you are using Motion Study, I'd turn off (disable) all contact sets.

Here is a screencast showing how I was able to get good behavior in this model by switching off both contact sets:

Jeff

[arnou-verfaillie]

@jeff_strater

Oh ok, so basically never use contact sets? I used em to see if the robot would be able to reach certain positions without hitting itself.

Thanks!

[arnou-verfaillie]

@jeff_strater

Thanks for the tip! It works fine now!
Here's a quick little video of the motions I wanted to make. I can only choose standard quality tho, not final with current gridsize? Is that because I'm using student version?

[jeff_strater]

I would not say "never use contact sets", but they will have this effect with Motion Studies.  I would more say "use them sparingly and carefully".  Another approach to the interference problem is to use an interference analysis, but that can be clumsy.

Jeff