Today while waiting for some image processing stuff, I worked on the project. Firstly, I implemented before and after state rewards for the parameters to add in more info as to how good or bad the parameter is. Whether this will be a good thing or not, I have yet to see (currently running extended game to see results).
MEGA PROBLEM: Forgot to put a negative reward when the parameter loses. Cancelling run as its results will be useless anyway.
Also, tried re-implementing the emergency (or panic) play mode which activates when the field is half full. Basically, when the field becomes half-full, the game is different, and the parameters that worked at a low field, may not work for a full field.
Doesn’t seem to work at the moment, but I’ll work on it some more later.