Spaces:
Running
Running
File size: 3,396 Bytes
55fed6d 812398b 55fed6d 812398b 55fed6d 812398b 55fed6d 812398b | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 | ---
title: Grid World Env
emoji: 🌍
colorFrom: blue
colorTo: green
sdk: docker
app_port: 8000
pinned: false
tags:
- openenv-0.10.0
- openenv
---
# Grid World Environment
A deterministic 5x5 grid world. The agent starts in the top-left corner and must reach the goal in the bottom-right corner. It's a small testbed for RL agents and a compact example of an OpenEnv environment (models, environment, server and client in a few files).
*The `/web` playground after a few steps: the agent, the goal and one button per direction.*
## Quick Start
Start the server from `envs/grid_world_env/`:
```bash
uv run server
```
Then connect with the client:
```python
from grid_world_env import GridWorldAction, GridWorldEnv
with GridWorldEnv(base_url="http://localhost:8000").sync() as env:
result = env.reset()
print(result.observation.message) # Welcome to Grid World! Goal is at [4, 4].
for move in ["DOWN"] * 4 + ["RIGHT"] * 4:
result = env.step(GridWorldAction(action=move))
print(move, result.observation.x, result.observation.y, result.reward, result.done)
```
The last step reaches `(4, 4)` with reward `1.0` and `done=True`. The API docs are at `http://localhost:8000/docs`.
## Docker
The image builds on the OpenEnv base image. From the repository root:
```bash
docker build -t envtorch-base:latest -f src/openenv/core/containers/images/Dockerfile .
docker build -t grid-world-env:latest -f envs/grid_world_env/server/Dockerfile .
docker run -p 8000:8000 grid-world-env:latest
```
`./envs/grid_world_env/test_grid_world.sh` (run from the repository root) builds both images, starts the container, checks the endpoints with `curl` and cleans up.
## Environment Details
### Action
**GridWorldAction**: `action`, one of `UP`, `DOWN`, `LEFT`, `RIGHT` (`MoveAction` enum or the string). `UP` and `DOWN` change `x`, `LEFT` and `RIGHT` change `y`. A move off the grid leaves the agent in place.
### Observation
**GridWorldObservation**:
- `x`, `y`: agent position, from `(0, 0)` to `(4, 4)`
- `message`: status message
- `reward`, `done`
`state()` returns the standard `State` (`episode_id`, `step_count`).
### Reward
- `-0.1` for every step that doesn't reach the goal
- `+1.0` for reaching the goal at `(4, 4)`, which ends the episode
There is no step limit: the episode only ends at the goal.
## Code Layout
- `models.py`: `MoveAction`, `GridWorldAction` and `GridWorldObservation`
- `server/grid_world_environment.py`: `GridWorldEnvironment` (reset, step, reward)
- `server/app.py`: the server, built with `create_app`
- `client.py`: `GridWorldEnv`, with a `step_move(MoveAction.UP)` helper
To build your own environment, see [Your First Environment](https://huggingface.co/docs/openenv/guides/first-environment).
## Hugging Face Space Deployment
This Space is built from OpenEnv environment `grid_world_env`.
- Space URL: `https://huggingface.co/spaces/openenv/grid_world_env`
- OpenEnv pinned ref: `0.10.0`
- Hub tag: `openenv`
### Connecting from Code
```bash
pip install git+https://huggingface.co/spaces/openenv/grid_world_env
```
```python
from grid_world_env import GridWorldEnv
with GridWorldEnv(base_url="https://openenv-grid-world-env.hf.space").sync() as env:
result = env.reset()
```
|