I am not too certain if this is possible, but I would like to replace GAZEBO physics with my own c codes.
Is it possible to get GAZEBO to accept acceleration, velocities and pose values such that it acts as a visualiser in the world created without using GAZEBO physics (ODE/Bullet)?
Welcome to the Community!
I know it’s possible to modify to source code to your taste, but I don’t know about replacing the physics with your own code. I suppose Gazebo can offer a better advise on that.
Please try https://answers.gazebosim.org/ and https://community.gazebosim.org/.
If you only want to vizualize your robot, use rviz
Thanks for your inputs.
I would like to continue to use GAZEBO’s simulation of sensors (cameras etc.) and the world environment, while using a different numerical simulator. Would that be possible if I just used rviz?
Regarding your original question, you might want to give this a read:
Regarding your new question: Can you elaborate a bit more onto what exactly you want to do? Which aspect do you want to simulate yourself?
Rviz is not a simulator, think of it as a monitor or a GUI for ROS. For example you can take the urdf model luploaded to the ROS parameter server, combine it with the /tf topic, which contains the transformations for all the links in the urdf model, and have it displayed as the robot in rviz. Or you can take a sensortopic, which is being broadcasted from either a real sensor or gazebo, and display that.
It however is not a tool to generate things.
Thanks a lot for your reply. I think it has pointed me to some resources I did not previously find.
I have a vehicle model dynamics and kinematics equations. Given certain actuator forces, the equations will be able to solve the velocities and pose of the AUV. This is similar to what GAZEBO does, but GAZEBO’s output is significantly different from my model, so I was keen on replacing it.
I think one way is to model my vehicle as an actor, and use my vehicle model to publish the velocities and pose to GAZEBO instead.
If you don’t mind me asking more questions, do you happen to know if it’s possible to publish a rostopic that GAZEBO can take in to update an actor’s pose? Also, if a sensor can be attached to an actor / be defined as an actor.