What we are doing is for each individual link, we split the link up into multiple links but only one needs "all of the above," the rest only need to look good. For example for our Upper Arm Link. The first link has a collision model which is a model for all the Upper Arm Links, inertial parameters which are the parameters for the entire upper arm, but then a visual that is only one small piece of the model. Then there are multiple additional links which only have require visuals that are the rest of the pieces of the model, but we had to put in minuscule inertial parameters and collision models in order to get that visual to come up. We are looking for a way for these additional links which only require a visual to only have a visual. Yes it is URDF, and yes the split up meshes are connected to the "main mesh," the one with inertia and collision, with fixed joints.
Rorry
Author: John Hsu Date: 2011-01-04 22:03 -500 To: User discussions Subject: Re: [ros-users] Gazebo Models
Hi Rorry,
Couple of questions, is it important that the robot
- has the right dynamics behavior,
- models collisions with objects,
- looks good in visualization or
- all of the above?
Is the current robot file written in URDF?
If yes, are you connecting the split up meshes with fixed joints?
John
On Tue, Jan 4, 2011 at 2:42 PM, Rorry Brenner
<rorry.brenner@resquared.com>wrote:
> We are trying to get a good visual for our robot in gazebo. Because of the
> built in gazebo smoothing we were not able to get our model to look right
> using the meshes we produced. As a workaround for this we split our meshes
> up into over 100 different files since gazebo does not smooth separate
> meshes together. The new model we produced looks great. However it runs
> extremely slow. We believe this is because of all the collision and physics
> calculations for the new links, but if we do not provide all of these
> additional links with inertial parameters and collision models they are not
> visualized in gazebo. I was wondering if there is a way to have gazebo
> still produce a visual if the link has no inertial parameters and no
> collision model?
>
> Thanks,
> Rorry
>
> _______________________________________________
> ros-users mailing list
> ros-users@code.ros.org > https://code.ros.org/mailman/listinfo/ros-users >
>