jitter in joints

A place to discuss everything related to Newton Dynamics.

Moderators: Sascha Willems, walaber

jitter in joints

Postby blackbird_dream » Mon Oct 22, 2018 10:35 am

Sorry to come back with recurrent issues.
I'm trying to separate sources of instabilities.
I've built a simple articulated mast fixed to the ground.
it comprises 4 dofs: tilt;lift;shift;expand
they are governed by silder_actuators and hinge_actuators.
The controls are defined in truck.cs script:
up/down arrow -> lift
F1/F2 -> tilt
F3/F4 -> expand
4/6 numpad -> shift

This simple test shows 2 issues:

1. if I don't add:
Code: Select all
nbm_f.SleepState = false;
in the script to update the joint actuators then the whole structure freezes in sleepstate, i.e. the input keys have no effect and the simulation's frozen after a while.

2. all the joints are stable except the forks which are jittering (expand and/or shift)

the model here:
https://we.tl/t-FFtxLE4ye3

the video here:
https://youtu.be/d7UROJSZFok
User avatar
blackbird_dream
 
Posts: 354
Joined: Wed Jun 07, 2006 3:08 pm
Location: France

Re: jitter in joints

Postby Julio Jerez » Mon Oct 22, 2018 11:36 am

ah this test does not add any extra force.

the it should work perfectly.

you say it does go to rest, that means the sole does find an equilibrium point, the problem has to be in the joints.

is the demo in the download?
Julio Jerez
Moderator
Moderator
 
Posts: 12249
Joined: Sun Sep 14, 2003 2:18 pm
Location: Los Angeles

Re: jitter in joints

Postby blackbird_dream » Mon Oct 22, 2018 12:42 pm

There is no extra force. The top structure hierarchically speaking is clamped to the ground as it connected to nothing.
the link is a winrar archive including the whole model.
Last edited by blackbird_dream on Tue Oct 23, 2018 5:53 am, edited 1 time in total.
User avatar
blackbird_dream
 
Posts: 354
Joined: Wed Jun 07, 2006 3:08 pm
Location: France

Re: jitter in joints

Postby blackbird_dream » Tue Oct 23, 2018 12:54 am

Isn't it a question of conflict between actuating forces and coulomb friction?

for better understanding here is an exploded view:
Image
User avatar
blackbird_dream
 
Posts: 354
Joined: Wed Jun 07, 2006 3:08 pm
Location: France

Re: jitter in joints

Postby blackbird_dream » Tue Oct 23, 2018 8:27 am

Here I removed the fork joint and I connected the colliders to part4. Then I update their position and orientation in realtime like this:
Code: Select all
     
        coll_fd = gm_fd.GetComponent<NewtonBoxCollider>();
        coll_fg = gm_fg.GetComponent<NewtonBoxCollider>();
        if (nbroot.ExpandPlus)
        {
            coll_fg.m_posit.x=coll_fg.m_posit.x+0.001f;   
            coll_fd.m_posit.x=coll_fd.m_posit.x-0.001f; 
        }
        else
        if (nbroot.ExpandMoins)
        {
            coll_fg.m_posit.x=coll_fg.m_posit.x-0.001f; 
            coll_fd.m_posit.x=coll_fd.m_posit.x+0.001f;   
       }

here is the video:

https://youtu.be/HVwMoO2V8kM
It doesn't jitter anymore but it seems the collision is not well detected between the load and the fork. Although the collider outlines moves, the collision is not actually updated.
For instance if I expand the fork too much the load doesn't fall inbetween the forks
Image
Maybe There is an instruction to actually activate the new position of the collider ? I know I'm skiing off-piste there.
User avatar
blackbird_dream
 
Posts: 354
Joined: Wed Jun 07, 2006 3:08 pm
Location: France

Re: jitter in joints

Postby Julio Jerez » Tue Oct 23, 2018 11:02 am

I understood correctly you say that the load is what is making it jitter?
some how if you remove the proble goes a way?
that could be a good cause, I though we has that fixed the was and option that solve the contact together with the joints. maybe is not set of they is a bug, I need to check.

I do not have time now, I am going to debug this over the weekend, but tell me you said you removed the load from the fork is does not jitter?
In the video I see going up and down, hwo do you move.
Julio Jerez
Moderator
Moderator
 
Posts: 12249
Joined: Sun Sep 14, 2003 2:18 pm
Location: Los Angeles

Re: jitter in joints

Postby blackbird_dream » Tue Oct 23, 2018 11:08 am

Yes. The load is making it jitter. It s heavy but not that enormous . 1205 kg.
With ligther weigths the pb vanishes. I can make a video if you wish.
User avatar
blackbird_dream
 
Posts: 354
Joined: Wed Jun 07, 2006 3:08 pm
Location: France

Re: jitter in joints

Postby Julio Jerez » Tue Oct 23, 2018 11:24 am

No, I think I have the info I need, I see if I can recreate it tonight in the sandbox.

There is a demo that is similar to what you are doing, I can use that to test the problem and if it work It rule out the possibility of an option no been enabled, if is happen then I see how to fix it.
if that is the problem, I make perfect sence, basically you can the entire island has to be send to the exact solver.

can you list the mass of the bodes involved, so that I use those values?
Julio Jerez
Moderator
Moderator
 
Posts: 12249
Joined: Sun Sep 14, 2003 2:18 pm
Location: Los Angeles

Re: jitter in joints

Postby blackbird_dream » Tue Oct 23, 2018 12:14 pm

I sent you the list by PM. I'll check tomorrow that the data are correct. Should be.
User avatar
blackbird_dream
 
Posts: 354
Joined: Wed Jun 07, 2006 3:08 pm
Location: France

Re: jitter in joints

Postby blackbird_dream » Wed Oct 24, 2018 2:43 am

they are correct except translator: Izz=7.4 but it doesn't change anything.
User avatar
blackbird_dream
 
Posts: 354
Joined: Wed Jun 07, 2006 3:08 pm
Location: France

Re: jitter in joints

Postby blackbird_dream » Tue Oct 30, 2018 11:25 am

Hi
have you succeeded in replicating the case study in the sandbox?
User avatar
blackbird_dream
 
Posts: 354
Joined: Wed Jun 07, 2006 3:08 pm
Location: France

Re: jitter in joints

Postby Julio Jerez » Tue Oct 30, 2018 11:29 am

not I was fixing some other older bugs, and merging the cmake script from Arpio.

I will try to do it sometime today and tomorrow.
Julio Jerez
Moderator
Moderator
 
Posts: 12249
Joined: Sun Sep 14, 2003 2:18 pm
Location: Los Angeles

Re: jitter in joints

Postby blackbird_dream » Tue Oct 30, 2018 11:58 am

Ok thk you very much
User avatar
blackbird_dream
 
Posts: 354
Joined: Wed Jun 07, 2006 3:08 pm
Location: France

Re: jitter in joints

Postby Julio Jerez » Sun Nov 04, 2018 12:38 pm

I was try to read the masses of the bodies by when I click the link I get this
Transfer deleted
Sorry, this transfer has been deleted and is not available any more


It does not matter, I think I can recreated, on my side.
I modified the forklift demo, in the sandbox by adding a compound that can be lifted.

when I made the weight 100 kg or left, is rock solid, not jitter at all,
I kept increasing and I see a minor jitter but remind solid until it hit about 400 kg.
where I start to see some estrange popping
I set the weight top 1000 kg, and it definably wrong the pop turns into an explosion.


before I move on to fix it, I like you to run the sand Box Demo and play with the the weight
let us say 200 kg and tell me is thsi is acceptable to you,
what I am try to get at is that we agree in what is acceptable and what is not. we may be seeing different thing.

The code is commited set to that demo, and the file you nee to look at is
../applications\demosSandbox\sdkDemos\demos\ServoJoints.cpp line 897


The should be eassy to get fix, in case you wonder why I do not go and made fix want and for all liek 1.53, the answer is that it is an engineering problem, I can make full fix at the expense of performance, but that will cause many people who are not particular interested on this kind of simulation reject teh engine, my solution is mak optional.

teh problem is that contacts can't be factored in a way the can be solve fast, so I have to be selective as to what contact can be made part of the contraction.
and those aren't at this time. therefore all in all it is remarkable that is can lift a 500 kg weigh and be stable.

anyway if you can sync an run that test, mean time I will add the fix.
Julio Jerez
Moderator
Moderator
 
Posts: 12249
Joined: Sun Sep 14, 2003 2:18 pm
Location: Los Angeles

Re: jitter in joints

Postby blackbird_dream » Mon Nov 05, 2018 8:32 am

i understand the fix is so severe it can impair speed performances.
Our concern is to lift up to 2500-3000 kg in the configuration uploaded.
With v1.5.3 it was stable but quite loose. Here it is solid (very good thing) but jittering (no jitter if I remove the expansion dof of the fork).
Ideally would be solid and stable of course. But if there is a param I can tune that would be fine.
With the unity demo (here :https://we.tl/t-olTmQqpQ5D)
it jitters with 2000kg and with 3000 kg the load is thrown away

Part1
M=203
body origin: X=0 Y=0 Z=0 (hinge axis)
CoM : X=0 Y=0.89 Z=0.226
Ixx=88.07
Iyy=13.98
Izz=99.94

Part2
M=177.75
body origin: X=0 Y=0 Z=0
CoM : X=0 Y=0.847 Z=0.215
Position : X=0 Y=h Z=0 (h=0.1)
Ixx=62.78
Iyy=17.07
Izz=77.31

Part3
M=101.21
body origin: X=0 Y=0 Z=0
CoM : X=0 Y=0.091 Z=0.338
Position : X=0 Y=h Z=0 (h=0.1)
Ixx=3.06
Iyy=7.36
Izz=9.77

Part4
M=58
body origin: X=0 Y=0 Z=0
CoM : X=0 Y=0.1 Z=0.411
Position : X=0 Y=h Z=0 (h=0.1)
Ixx=1.7
Iyy=5.8
Izz=7.4

Part5
M=46.6
body origin : X=0 Y=0 Z=0 (middle of extremities)
CoM : X=d(0.6) Y=0.148 Z=0.707
Position : X=0 Y=h Z=0 (h=0.1)
Ixx=5.9
Iyy=4
Izz=1.9

Part6
M=46.6
body origin: X=0 Y=0 Z=0 (middle of extremities)
CoM : X=-d Y=0.148 Z=0.707
Position : X=0 Y=h Z=0 (h=0.1)
Ixx=5.9
Iyy=4
Izz=1.9

h:lift height
User avatar
blackbird_dream
 
Posts: 354
Joined: Wed Jun 07, 2006 3:08 pm
Location: France

Next

Return to General Discussion

Who is online

Users browsing this forum: No registered users and 17 guests

cron