This repository contains Python scripts that demonstrate how to use Plant Simulation as a learning environment for Reinforcement Learning (RL) algorithms to tackle deadlock situations in Automated Guided Vehicle (AGV) systems. The code is compatible with the Gymnasium and Ray libraries.
This file is the main entry point for training and testing multi-agent reinforcement learning models using Ray RLlib.
This is a single-agent environment that simulates an AGV system with deadlock capabilities. It is implemented using the Gymnasium framework.
This is a multi-agent environment that simulates an AGV system with deadlock capabilities. It is implemented using the Gymnasium framework and is compatible with Ray's MultiAgentEnv.
This file contains another multi-agent environment for AGV systems, implemented using the Gymnasium framework and compatible with Ray's MultiAgentEnv.
This directory contains various simulation model files used by the environments.
This directory is used to store experiment data, including logs and results.
- Python 3.x
- Ray 2.35.0 (recommended)
- Gymnasium
- Plant Simulation > 2201
- PyTorch (for CUDA testing)
PS: Sorry, everything is still messy, I hopefully will organize it soon.