Hey,
I'm using Argos to learn path planning by machine learning. For this I'm using the Foot-Bot with the dynamics2d physics engine in simulation, but in reality I'm using the TurtleBot3. I got most of it working, but I've got one problem: the learning algorithm learns to rotate the robot very much, which seems to work good for the Foot-Bot but not for the TurtleBot3, because he rotates far too much for the same rotation command. After looking at many things and thinking about it, my guess is that the robot in simulation has a lower acceleration than the TurtleBot3. So I went searching where exactly the acceleration is defined in the simulation and for this I searched where the motion is applied to the robot, but after nearly 2 days of searching I didn't really find it. I just found that it probably happens when the CPhysicsModel::UpdateFromEntityStatus gets called by CDynamics2DEngine::Update, but I don't really know where.
So my question is: Where does the motion application happen exactly and is there a way to change the acceleration in the simulation or can I at least get the acceleration somewhere to adapt the commands sent to the TurtleBot3 in reality? Or is my only possibility to model the TurtleBot3 robot in the simulation?
Thank you very much in advance!