Module `Fehu_envs`

Built-in environments for testing and learning.

Four environments covering the standard RL benchmarks: a simple 1D walk, the classic cart-pole, a grid navigation problem, and the sparse-reward mountain car. All follow the Fehu.Env interface.

Environments

module Random_walk : sig ... end

One-dimensional random walk.

module Cartpole : sig ... end

Classic cart-pole balancing (CartPole-v1).

module Grid_world : sig ... end

5x5 grid navigation with obstacle.

module Mountain_car : sig ... end

Mountain car with sparse reward (MountainCar-v0).