I am working on a project that involves modifying Mujoco environments for reinforcement learning. I want to create a new environment that is similar to the InvertedPendulum environment, but with a different initial state and change it from termination to truncation like the Pendulum environment. Instead of starting with the pendulum upright, I want to start with the pendulum hanging down and the cart in the center, and try to swing it up to the upright position. I would also like to be able to change the parameters of each part because I am going to implement this in real life for my thesis.
How can I modify the XML file and the Python code of the InvertedPendulum-v4 environment to achieve this? Any help or guidance would be appreciated.