My wheeled robot model sometimes flies up into the air while driving. Why?
I'm new to Gazebo, so I probably just have something set incorrectly. But many of these robot model files (SDF) are made by hand using examples and tutorials found on the Internet, so they are quite error prone, I think.
Which SDF file properties will cause a wheeled robot to break loose from the ground plane and take off flying straight up into the air at extreme velocity? Do I have my friction values in the wheels messed up? Or is it the chassis or wheel link inertia values that I need to be more mindful of?
It's a 4-wheel skid steer platform (that is not meant to fly), so it is using the skid steer plugin. If the torque on the skid steer plugin is powerful and if the wheel friction values are incorrect, could it bind and then break loose and snap the entire model up into the air?
I'm using Gazebo 1.9 and ROS Hydro.
Asked by Kurt Leucht on 2014-07-03 14:59:02 UTC
Answers
you can try setting max_vel
parameter to 0, e.g.:
<collision name="...">
<surface>
<contact>
<ode>
<max_vel>0</max_vel>
</ode>
</contact>
</surface>
</collision>
this will prevent/reduce interpenetration error from injecting energy into the system.
Asked by hsu on 2014-07-21 11:49:33 UTC
Comments
Per this question: http://answers.gazebosim.org/question/6584/my-mobile-robot-slowly-rolls-with-no-forces/
... I had some instability problems with my robot model definition which caused it to slowly roll on the ground plane with no forces applied. Since I got that issue solved, I have had no more instances of my robot flying up into the air. So the two problems were probably due to the same model definition issue. Buggy model, in other words. Thanks.
Asked by Kurt Leucht on 2014-08-11 19:38:16 UTC
Comments
Some of these links are damaged. This one is. PLease could you point us to the current functional link if you don't mind? ThANKS.
Asked by afroRoboticist on 2019-03-30 21:25:58 UTC
Yeah, that link is broken. Not my fault. It worked in August of 2014.
Asked by Kurt Leucht on 2020-04-03 12:58:48 UTC
Comments