Hi Jim,
The reason your spring damper setup is unstable is because the service calls
in your setup is not fast enough to update every simulation cycle.  Just for
the sake of experimentation, if you slowed down simulation by throttling
simulation update rate, adding:
  <updateRate>100.0</updateRate>
to <physics:ode> block in empty.world, then the spring damper model should
work correctly.  This is not the right solution, ultimately, it is better to
model the spring damper system as a transmission or using gazebo plugins;
this way, the forces are updated at every simulation time step.  Let's chat
off-line if you want to talk about implementation.
hope this helps.
John


On Tue, Feb 1, 2011 at 7:51 PM, Jim Rothrock <
jim.rothrock@wunderkammerlab.com> wrote:

> I am writing a package that simulates a Traxxas E-Maxx RC truck in Gazebo.
> Since springs (for the shock absorbers) cannot be specified in a URDF
> dynamics block, I've created a node that reads the shock absorber joint
> positions (they are prismatic joints), uses Hooke's law to compute spring
> forces, then applies the forces to the springs. The node also simulates
> damping. No matter what damping coefficient I use, the truck continues to
> oscillate. Sometimes it appears that Gazebo's ground contact force
> simulation is kicking the wheels into the air, but that's just a hypothesis.
> I've attached a test package to this message, and I'd really appreciate it
> if someone could take a look at it. To execute the test, run "roslaunch test
> test.launch" in one window, then run "rosrun test test.py" in another. The
> oscillating truck should be displayed in a Gazebo window.
>
> --
> Jim Rothrock | Wunderkammer Laboratory
> jim.rothrock@wunderkammerlab.com
>
>
> _______________________________________________
> ros-users mailing list
> ros-users@code.ros.org
> https://code.ros.org/mailman/listinfo/ros-users
>
>