[ODE] Combing setpos'ed "ghost" bodies with a rigid central core?Doable?

[Jon Watte]

> but would it be practical for me
> to setpos bodies around IF those bodies were considered 
> "ghosts" by the rest of the world?  (eg. they exist only as 
> sensors to detect collision, never adding contact joints themselves)

Why would you use a body if you don't want to create joints for 
them? Just create the dGeom and test against that, no body needed.

> My concern is that I'm not sure what sort of body position 
> persistence is going on behind the scenes - how exactly the 

You have the code; just read it! The answer is: not much. ODE 
uses a forward Euler integrator, which needs no state other 
than the publicly visible position and velocities.


The other alternative is to actually use bodies for the main 
pieces of arms and legs, and apply forces in a "chase" mode to 
drive them using forward kinematics. It can work pretty well; 
especially if you use "predictive chase": looking at the 
animation, where should the arm be two steps from now? Looking 
at current angular/linear velocity, where will it be two steps 
from now if nothing changes? Apply a force to make the 
appropriate velocity change, clamped by whatever your max 
acceptable force is.

Cheers,

			/ h+