Getting Started

Welcome to custom_grid_env! This page will help you get familiar with the environment quickly.

Concepts

The environment simulates a grid where an agent must navigate. The following components play a central role:

1. AgentInterface: The primary interface for your AI agents. It encapsulates the environment, the ghost, and the particle filter.
2. Particle Filter: A mechanism for estimating the agent's position if it is not exactly known (localization).
3. CNN Classification: A neural network that processes images of grid cells to recognize objects like dogs or flowers.

First Experiments

Notebook	Description	Link
Interactive GUI	Full dashboard with sensors and PF visualization.
Environment Demo	Learn the basics of control.
CNN Training	Learn how to train the neural network that the agent uses for object recognition.

A Simple Example

Here is a minimal script to start an agent with random movements:

from custom_grid_env.interface import AgentInterface
from custom_grid_env.agents.random_player_agent import RandomPlayerAgent

# Initialize interface
interface = AgentInterface(render=True)
obs = interface.reset()

# Create agent
agent = RandomPlayerAgent(interface.get_action_space())

# Run episode
for _ in range(100):
    action = agent.get_action(obs)
    obs, reward, done, info = interface.step(action)
    if done:
        break

interface.close()

Further Tutorials

Check out our detailed tutorials: - CNN Training Tutorial - Particle Filter Tutorial