I am fairly new to RL and I compete in AWS DeepRacer student league. The main task there is to create a reward function. All the hyperparameters and action space are fixed. So far, I know how to shape a single component like speed or steering. I always do like steer_reward^2 or speed^2. But when I try to combine the components together, I am not seeing good results. Is it best to mutliply or add components and are there any resource where I find various maximization functions?
Asked
Active
Viewed 176 times
0
-
Hey, check these answers: one, two. Did they answer your question? – Aray Karjauv Sep 06 '22 at 13:03
-
@ArayKarjauv Nope. I am trying find out functions that have a good global minimum or maximum where the variables of the functions are the speed or steering or other params. – aniketvp24 Sep 11 '22 at 19:28