[ODE] Geoms vs. Joint connected bodies

Daniel Hein dhein at informatik.hu-berlin.de
Tue May 2 17:38:45 MST 2006


I have an articulated body (robots arm), consisting of 3 boxes, each 
with different weight. This body is connected via a hinge joint(1) to 
the (static) world. In my simulation I encountered different behavior, 

- arm is built as one body, consisting of 3 geoms, individual box masses 
are translated accordingly.
- arm is built from 3 different boxes, connected by 2 hinge joints in a 
row; (these hingejoints are fixed by AMotors; the axes of these joints 
are perpendicular to joints(1) axis)

The total mass of both constructions are completely equal, including 
mass, center of mass and inertia matrix - I already compared.

But if I add continuously a little constant torque to the joint(1), the 
simulation results differ a lot: e.g. meanwhile with the first 
construcion method the arm begins to rotate, with the second 
construction it just oszillates around the zero hanging. Decreasing 
dWorldStep (act: 0.01) does not affect. For having more or less same 
results I have to add to the second construction nearly the twice of the 
torques of the first construction.(?) Same completely different behavior 
with the use of angular motors (fmax/vel). I do not use any of ODEs 
optimization options.

Is this a known behavior? Besides the mass properties, how else the arms 
could differ, causing these differences?

If wished, I could extract/prepare a simple code fragment showing this 

Thanks in advance,

Daniel Hein

More information about the ODE mailing list