I also made so that if it is not learning anything, it will retrace its steps.
So, every time it hits a wall, it has to start learning all over again?
Mine does that to...
It deletes the past two moves if its score decreases from the last generation, allowing it to essentially rethink its moves, that is how mine evolves.
Same, but it removes more than 2, making the learning process faster.
I just don't want to go overboard and erase tons of potential progress.
Mine remove the previous 15 steps to keep the bot from stalling. If you remove too little, it can get stuck because it cant get back out of a stuck spot.
I used to only delete one, and the stalling issue is why I made it two. Maybe I'll take your advice and see what changes..
Welp, experimented a lot and there is nothing I can do to make it much better, there have been quite a bit of changes in the code, and I think it works worse than it did before.. sad.
Did you know "recover exists?"
Nope, how do I do that.
nevermind, found it.
Edit: Anddd the original is gone, only the experimental ones exist now in the recover. the only one left is not good.
I think I have a remix...
No, I don't...
It's okay to be sad (it's also okay to feel angry feelings or similar), when a lot of effort was put into something, but it doesn't work the way it was hoped it would. However, you probably have still learnt something in the process.
So, if learning and having fun are two of your goals, then you succeeded in those two goals (well as long as you don't push yourself too much so that it ceases to be fun for you), no matter what happens to the project you were working on. Agree?
I suppose you are right, I did see some cool things so I am completely fine with this end result. I may make a bot with a similar algorithm to do something else.