Simplified dynamics model vs URDF + Gazebo
I try to simplify the explanation, but feel free to comment if something is unclear.
I have few remote guided mobile robots (rovers like this one) and a Ground Station for the communication of their information (from onboard sensors) among them and to the PC.
I'm going to build a completely distributed autonomous swarm system where all the robots are managed by distinct ROS nodes running on the PC (at the moment I can't directly handle it onboard).
I'd like to implement also simulated agents in the swarm. For this purpose I have a simplified model of the rover dynamics and I could use it for the simulation. But if I'd like to represent the entire system (e.g. in Gazebo) will it be adequate to create just the URDF model of the robot (which, together with Gazebo, implements the robot dynamics)? Or should I associate the simplified dynamics to the URDF model in some way?
I think that the URDF approach should be preferred to a simple C++ implementation with a simplified dynamics. Is it right? Or could it be too much computationally expensive and difficult to implement?
And last but not least, how to create a URDF model from scratch? I mean, what is important to include for a mobile robot like the previous one? What can be approximated?
