https://github.com/nonkloq/mazeharvest

A Grid Based RL Environment & Implementaions of few Deep-RL Algorithms.
https://github.com/nonkloq/mazeharvest

dqn-pytorch ppo-pytorch pytorch reinforcement-learning rl-algorithms-pytorch rl-environment

Last synced: 2 months ago
JSON representation

A Grid Based RL Environment & Implementaions of few Deep-RL Algorithms.

• Host: GitHub
• URL: https://github.com/nonkloq/mazeharvest
• Owner: nonkloq
• Created: 2025-01-12T09:15:35.000Z (4 months ago)
• Default Branch: main
• Last Pushed: 2025-01-29T15:23:26.000Z (4 months ago)
• Last Synced: 2025-01-29T16:28:12.790Z (4 months ago)
• Topics: dqn-pytorch, ppo-pytorch, pytorch, reinforcement-learning, rl-algorithms-pytorch, rl-environment
• Language: Python
• Homepage:
• Size: 32.9 MB
• Stars: 0
• Watchers: 2
• Forks: 0
• Open Issues: 0
• Metadata Files:
  • Readme: README.md

Awesome Lists containing this project

README

        
# MazeHarvest

MazeHarvest is a partially observable, stochastic reinforcement learning environment. For a detailed description, refer to [homegym/README.md](./homegym/README.md). In summary, the agent must rely on sensory data like ray perception and heuristics to accomplish the following tasks:

- **Kill hostile moles** that hunt the agent.

- **Harvest toxic plants** to reduce environmental toxicity and survive.

- **Navigate complex mazes** in advanced settings.

To install `homegym`, clone this repository and follow the instructions in [homegym/README.md](./homegym/README.md).



  Agent Details

  > RDQN Agent, CNN Policy. No Episode Memory (State In Action Out).

---

## Experiments 

I have Implemented few RL algorithms using Pytorch in the [experiments/algos](./experiments/algos) directory to solve this environment. Each folder contains these two file:

- **toy_test.py**: Inherits the base algorithm class to test it on gymnasium environments and logs the test results in `ttresults.md`.

- **ttresults.md**: Logs with:

    Hyperparameters used during training.

    Descriptions of the algorithm and its implementation.

    Test results: performance across different environments.

    Evaluation scores for 100 episodes.

**Test environments**: The algorithms are tested on Gymnasium toy environments to ensure the implmentation robustness.

### Implemented Algorithms

| Algorithm     | Folder         | Description                                                                 |

|---------------|----------------|-----------------------------------------------------------------------------|

| **PPO**       | `algos/ppo`    | Simple implementation of Proximal Policy Optimization (PPO) with Generalized Advantage Estimation (GAE). |

| **Rainbow DQN** | `algos/rdqn`  | Combines C51, N-Step Learning (random M to N steps, instead of fixed N steps), Prioritized Experience Replay (PER), Double Q-Learning, and Dueling Network Architecture. |

---

### Utilities

The [experiments/eutils](./experiments/eutils) directory contains utility functions for:

- Processing observations.

- Training & Testing models for this environment.

---

### Training Scripts

Each training script follows the naming convention `*_train.py`, where `*` corresponds to the algorithm name. For example, `ppo_train.py` is the script to train the MazeHarvest environment using PPO.