I have done some reasearch on Q learning but all I find is Python explanations of it and I need help with an experiment I am trying to do.
Basically I want a pathfinding AI that gets rewards if it reaches a player in 30 seconds or less but gets punished if it didn’t reach the player in time in a maze. My inputs are Player Position, Enviormental terrain(maze raycasting), and Humanoid Walkspeed which is capped at 30. My problem is I am just starting to learn Algebra I in school next year and I have no clue on how this algorithem trains the ai’s neural network. Use the basic algorithem below:
If you can help me.
If I were you I’d scrap the idea of using Q-learning or any other machine learning technique for your pathfinding AI. There are many great (and much easier) pathfinding algorithms out there. Q-learning would just add more time to your development process to get you the same result as if you used a different algorithm.
Here’s a video visualizing several pathfinding techniques that you could implement instead.
1 Like
Yes, I know pathfinding is a way eaisier way to code that, but I want to do things more advanced such as Q Learning but I want to figure out how Q Learning works. Picture this, if you touch a wall inside of a maze, that is randomly generated, then you die(punishment).Also if you get to the end you get a reward(reward). I could use pathfinding, but I want my AI to learn where to go using some sort of machine learning. It doesn’t have to be the Q-Learning algorithm but I do want it to be machine learning.
If you want to learn how to use Q-learning or machine learning than I would suggest using the numerous online resources available rather than asking a question on the Roblox DevForum about it. No offense to this forum, but it is definitely lacking in “advanced mathematicians” or programmers who understand anything about machine learning. You could still (and should) use Roblox and luau as a testbed for your explorations, but as far as general machine learning knowledge goes, use google instead.
Hope this helps and good luck 
.
2 Likes
Ok, I will look and continue the path for q learning and machine learning. I will probably post something on stack overflow to see if anyone knows anything there. Thanks tor your help.
1 Like
