Help me, please. This is normal behaviour of the train, when, during training, he was going off the rails? I think I read somewhere that the train should remain in place if the agent commits an unacceptable action
I think this is just a rendering issue. Looking at the observation rendering in the image you posted i think the render did not redraw the new rails.
Can you please try to do
env_renderer.reset() where env_renderer is the renderer you defined.
Does this resolve your issue?
What version of FLATland are you using - it might be a bug which was reported and fixed few days ago.
Thanks, Erik. It works!