The OM_description doesn't work well with Gazebo

[Yiping-Steven]

1. The rviz and the gazebo loaded our robot_description in different ways. Our implementation of the robot_description doesn’t work well with gazebo. I feel the problem could be related to the OM-gazebo plugin, but this is just my intuition, I could be wrong.
2. The current “weird” model of OM in gazebo, doesn’t have a sense of collision. It could be either there’s no collision tag in OM description at all, or there IS the collision tag but it’s not interpreted correctly by gazebo. You need to check that, if it’s the latter case, then my guess of the 1st issue is more probable.

[Yiping-Steven]

After yesterday's good work, the collision and the floating issue of the OM model in gazebo is solved, but the new issue is the OM model controller has some strange dynamic behavior and will interfere with the navigation controller, even when itself is in silence (not receiving any command). Two way to solve this:

1. One easy-fix: Fix all the joints ( make sure they are in good zero position first), take out the gazebo plugin controller. Make the OM a rigid body attached to turtulebot. This way is proved to work, but has an obvious disadvantage of discarding all the manipulator's capabilities out of this simulation. So we need to create another simulation environment to compensate that. Specifically, we need to make another urdf, fix the turtlebot, only allow the OM to move and interact with the vision. Since Manipulation Module and the Navigation Module are pretty decoupled, so this can actually work.
2. Fix it the hard way: Look into what's wrong with the OM's gazebo plugin controller and figure out why it has such a strange interference on the navigational movement. And also why any tiny change in mass or inertia would make the gazebo model to fail.