If you think about it, this what reinforcement learning folks are doing. They use the vanilla state and then they lift it to the observed state by doing some additional calculation with the original state data.

For example you start with the raw coordinates of snake in a snake game, but you now can calculate how many escape routes the snake has, and train on it.

That’s correct

I’ve spent most of the last 20 years doing reinforcement learning and it is exceptionally simple conceptually

The challenge is data acquisition and the right types of process frameworks