Tempest_in_Action/notebooks/environments/Minigrid/minigrid/envs/fourrooms.py

from __future__ import annotations

from minigrid.core.grid import Grid
from minigrid.core.mission import MissionSpace
from minigrid.core.world_object import Goal
from minigrid.minigrid_env import MiniGridEnv


class FourRoomsEnv(MiniGridEnv):

    """
    ## Description

    Classic four room reinforcement learning environment. The agent must
    navigate in a maze composed of four rooms interconnected by 4 gaps in the
    walls. To obtain a reward, the agent must reach the green goal square. Both
    the agent and the goal square are randomly placed in any of the four rooms.

    ## Mission Space

    "reach the goal"

    ## Action Space

    | Num | Name         | Action       |
    |-----|--------------|--------------|
    | 0   | left         | Turn left    |
    | 1   | right        | Turn right   |
    | 2   | forward      | Move forward |
    | 3   | pickup       | Unused       |
    | 4   | drop         | Unused       |
    | 5   | toggle       | Unused       |
    | 6   | done         | Unused       |

    ## Observation Encoding

    - Each tile is encoded as a 3 dimensional tuple:
        `(OBJECT_IDX, COLOR_IDX, STATE)`
    - `OBJECT_TO_IDX` and `COLOR_TO_IDX` mapping can be found in
        [minigrid/minigrid.py](minigrid/minigrid.py)
    - `STATE` refers to the door state with 0=open, 1=closed and 2=locked

    ## Rewards

    A reward of '1 - 0.9 * (step_count / max_steps)' is given for success, and '0' for failure.

    ## Termination

    The episode ends if any one of the following conditions is met:

    1. The agent reaches the goal.
    2. Timeout (see `max_steps`).

    ## Registered Configurations

    - `MiniGrid-FourRooms-v0`

    """

    def __init__(self, agent_pos=None, goal_pos=None, max_steps=100, **kwargs):
        self._agent_default_pos = agent_pos
        self._goal_default_pos = goal_pos

        self.size = 19
        mission_space = MissionSpace(mission_func=self._gen_mission)

        super().__init__(
            mission_space=mission_space,
            width=self.size,
            height=self.size,
            max_steps=max_steps,
            **kwargs,
        )

    @staticmethod
    def _gen_mission():
        return "reach the goal"

    def _gen_grid(self, width, height):
        # Create the grid
        self.grid = Grid(width, height)

        # Generate the surrounding walls
        self.grid.horz_wall(0, 0)
        self.grid.horz_wall(0, height - 1)
        self.grid.vert_wall(0, 0)
        self.grid.vert_wall(width - 1, 0)

        room_w = width // 2
        room_h = height // 2

        # For each row of rooms
        for j in range(0, 2):

            # For each column
            for i in range(0, 2):
                xL = i * room_w
                yT = j * room_h
                xR = xL + room_w
                yB = yT + room_h

                # Bottom wall and door
                if i + 1 < 2:
                    self.grid.vert_wall(xR, yT, room_h)
                    pos = (xR, self._rand_int(yT + 1, yB))
                    self.grid.set(*pos, None)

                # Bottom wall and door
                if j + 1 < 2:
                    self.grid.horz_wall(xL, yB, room_w)
                    pos = (self._rand_int(xL + 1, xR), yB)
                    self.grid.set(*pos, None)

        # Randomize the player start position and orientation
        if self._agent_default_pos is not None:
            self.agent_pos = self._agent_default_pos
            self.grid.set(*self._agent_default_pos, None)
            # assuming random start direction
            self.agent_dir = self._rand_int(0, 4)
        else:
            self.place_agent()

        if self._goal_default_pos is not None:
            goal = Goal()
            self.put_obj(goal, *self._goal_default_pos)
            goal.init_pos, goal.cur_pos = self._goal_default_pos
        else:
            self.place_obj(Goal())