Questions about runner.py, training process and purpose of anymalController_00000000.cpp

[habin0615]

what I understand..

For training our owns agent, we have to use training code like runner.py.

1. then we have to generate another code which can train our own agents like runner.py??

For agent to learn sumo action, reward function should be changed.
2. In runner.py, it looks like reward is coming from env, is that correct? I would like to know where reward is coming from and how I can modify it.

3. I'm not sure what the AnymalController_00000000.cpp code is supposed to do. I want to know what the main purpose of AnymalController_00000000.cpp is, and I want to know when and where to use it.