[ODE] movement force conversion

Matthias Baas baas at ira.uka.de
Tue Mar 15 11:07:40 MST 2005


Piotr Obrzut wrote:
> that is what interest me in longer time, but now before any
> research: it is possible to turn stored forward kinemtic animation
> into forward dynamics? 

Well, in principle, yes. From the kinematic animation you could compute 
the forces/torques that are necessary to do this particular movement 
("inverse dynamics" is the keyword here). After that, you could use 
forward dynamics and apply the previously computed forces and torques.
However, in practice, this is not as easy as it sounds. Blindly applying 
the forces/torques will most certainly result in a motion that quickly 
diverges from the original motion. So you have to incorporate some sort 
of control algorithm that counteracts this divergence.
Another thing is, that it's not always possible to uniquely determine 
the forces/torques. If there are closed kinematic loops you can't know 
the original forces that have been applied (for example, if you're 
holding a box with both hands (=kinematic loop) then you won't be able 
to compute the forces just by looking at the resulting motion. You can't 
"see" how much pressure you applied to the box).

- Matthias -


More information about the ODE mailing list