Trait Env

pub trait Env<Obs, Action> {
    // Required methods
    fn reset(&mut self);
    fn step(&mut self, action: &Action) -> (bool, f32);
    fn observe(&self, obs: &mut Obs);
}

Main trait implement by each environements

Required Methods

fn reset(&mut self)

Reset function to restart the environement

fn step(&mut self, action: &Action) -> (bool, f32)

Step function to take an action, return a tuple of (done, reward). Done will be true when the episode is finished and require to call .reset() next The reward is simply the reward signal.

fn observe(&self, obs: &mut Obs)

This method is used get the observation, this is the main difference with the OpenAI gym, the goal is to be able to reuse objects and avoid allocation..

Implementors