3 replies to this topic

[Vasco_F]

Hi,

I'm developing a video game, in which I'm making a charactar with rigid-body physics (also known as "ragdoll" physics). The way I've made it is probably not completely realistic, because I only use velocity vectors to calculate the position of each joint of the "ragdoll", based on an initial velocity vector applied to a joint.

If you want to check it out, you can download it here.

The way I do this is illustated in this image:



Anyway, the problem I have now is how to calculate the velocity vector of joints that make an angle that is restricted (an angle that shouldn't get any bigger, for example). This simplified diagram illustrates my problem:


In the diagram, how should I calculate vectors v1 and v2? Note that in the diagram, the entire body should be rigid, because of the angle restriction.

I would truly appreciante any help on this...

[Reedbeta]

I'm not sure this problem can be solved only using a velocity-based representation without forces. I would advise you to read this Gamasutra article, which is about ragdolls (requires free registration).

[Vasco_F]

Thanks, if I can't solve the problem this way I'll try that

[SigKILL]

Have you tried using straight forward lagrange multipliers? I´m not seeing any reason why this should not work in a velocity based system (typically can equation based algortihms be used either way). Since your ragdolls probably doesn´t containt loops you can look for baraff´s paper called "Linear-time dynamics using Lagrange multipliers" which provides a pseudo code that does this in linear time (in a force based system, but if you know your physics I believe it should be quite easy to convert those formulas).