# Room These tasks are set in a single square room, where the goal is to reach the staircase down. There are multiple variants of this level. There are two sizes of the room (`5x5, 15x15`). In the simplest variants, (`MiniHack-Room-5x5-v0` and `MiniHack-Room-15x15-v0`), the start and goal position are fixed. In the `MiniHack-Room-Random-5x5-v0` and `MiniHack-Room-Random-15x15-v0` tasks, the start and goal position are randomised. The rest of the variants add additional complexity to the randomised version of the environment by introducing monsters (`MiniHack-Room-Monster-5x5-v0` and `MiniHack-Room-Monster-15x15-v0`), teleportation traps (`MiniHack-Room-Trap-5x5-v0` and `MiniHack-Room-Trap-15x15-v0`), darkness (`MiniHack-Room-Dark-5x5-v0` and `MiniHack-Room-Dark-15x15-v0`), or all three combined (`MiniHack-Room-Ultimate-5x5-v0` and `MiniHack-Room-Ultimate-15x15-v0`). The agent can attack monsters by moving towards them when located in an adjacent grid cell. Stepping on a lava tile instantly kills the agent. When the room is dark, the agent can only observe adjacent grid cells. Examples of the `MiniHack-Room-Ultimate-15x15-v0` task: ![](../imgs/rooms.png) ## Reward The agent receives a reward of +1 for reaching the goal. ## Source [Source](https://github.com/facebookresearch/minihack/blob/main/minihack/envs/room.py) ## All Environments | Name | Capability | | --------------------------------- | -------------- | | `MiniHack-Room-5x5-v0` | Basic Learning | | `MiniHack-Room-15x15-v0` | Basic Learning | | `MiniHack-Room-Random-5x5-v0` | Basic Learning | | `MiniHack-Room-Random-15x15-v0` | Basic Learning | | `MiniHack-Room-Dark-5x5-v0` | Basic Learning | | `MiniHack-Room-Dark-15x15-v0` | Basic Learning | | `MiniHack-Room-Monster-5x5-v0` | Basic Learning | | `MiniHack-Room-Monster-15x15-v0` | Basic Learning | | `MiniHack-Room-Trap-5x5-v0` | Basic Learning | | `MiniHack-Room-Trap-15x15-v0` | Basic Learning | | `MiniHack-Room-Ultimate-5x5-v0` | Basic Learning | | `MiniHack-Room-Ultimate-15x15-v0` | Basic Learning |