Powered ragdolls

A place to discuss everything related to Newton Dynamics.

Moderators: Sascha Willems, walaber

Powered ragdolls

Postby JoeJ » Fri Feb 25, 2011 5:27 pm

I'm currently playing around with something i've failed years ago in my own simple physics engine: Powered ragdolls.
I setup joints as usual and try to get motion from setting torques, the joints do the linear part.
I don't want to trick the ragdoll to reach some worldspace target,
instead the motion should only come from reaching the target orientation offset between 2 bones.
This isn't easy, so my question is: Is something like this planned for future release?

And optional: Would it be better to use the joints itself as motors?
If so, what practices do you suggest, after some testing using SetRowSpringDamperAcceleration looks promising, but jitters.
Can you also give some explantation about SetRowM**imunFriction and SetRowAcceleration? Wiki and forum ar a little silent about it.
Maybe i can get it stable by changing those settings over time and in relation to the current error.
User avatar
JoeJ
 
Posts: 1489
Joined: Tue Dec 21, 2010 6:18 pm

Re: Powered ragdolls

Postby Julio Jerez » Fri Feb 25, 2011 6:36 pm

there is something like that in newton already it is the joint CustomKinematicController
basically you set a target matrix, or a target rotaion ot a target rotations and the jopint will align to the position.
It is what I use for objec picking in the demos, and the tutorials.

The problem is that you wan to do it by applin torque an dthat will never work like that.

The problem you ask is "what is the forze I need apply to this object so that is align to this other position?"
it is a calculation of a force or torque as a function of the relative position or relative angle.
what you can get is the force the engine calculated to achive the alignment, you cna even get the valuse before teh force is applied.
but you can not do is to applied a force and hope motion will achive the aligment, that will be very wrong and lead to the jitter you always have.
Julio Jerez
Moderator
Moderator
 
Posts: 12426
Joined: Sun Sep 14, 2003 2:18 pm
Location: Los Angeles

Re: Powered ragdolls

Postby JoeJ » Sat Feb 26, 2011 12:56 pm

But i guess, the powered joint approach worked for walabers ski demo?
Maybe he got the targets from worldspace oriented algo, so no relativity.

However, i've got the torque method working very, very well meanwhile.
stiffness and fast reaction comes from taking smaller timesteps (which i've expected and is ok),
and jitter disappears completely when i fake the bodies to have uniform inertia.

I'm sure i can get a walking ragdoll, if i'm able to develop the AI.
This will take a lot of time, but then i'll post a video...
User avatar
JoeJ
 
Posts: 1489
Joined: Tue Dec 21, 2010 6:18 pm


Return to General Discussion

Who is online

Users browsing this forum: No registered users and 1 guest

cron