Dear @Deathcalibur,
Thank you again for the plugin and all your hard work. I was hoping you could clarify something regarding the different types of observations (set, static arrays, struct…).
For instance, is using a struct observation containing three float values to represent a location equivalent to using a dedicated location observation? In this case, I’m not referring to the inverse transform, but rather how this distinction might affect the PPO context, as it seems to have an impact on the encoder architecture.
Thank you