RSS 2026 Submission: Interactive Knowledge Distillation with Adaptive Teachers in Cooperative Multi-Agent Reinforcement Learning

This repository contains the official implementation of our RSS 2026 paper based on the HINT (Interactive Knowledge Distillation with Adaptive Teachers in Cooperative Multi-Agent Reinforcement Learning) framework for cooperative multi-agent reinforcement learning (MARL).

Project Structure

├── environment.yml          # Conda environment file
├── HINT/                    # Source code for HINT training and evaluation
├── ocean/                   # Supporting replenishment-at-sea wave information
└── README.md                # Project instructions

Getting Started

Step 1: Set Up the Environment

You will need Anaconda installed.

Create and activate the conda environment:

conda env create -f environment.yml
conda activate ras_env

Step 2: Run the Main Experiment

Navigate to the implementation directory and launch the training script:

cd HINT
python HINT_20x20_MARINE_IT_DAgger_MF_Het.py

ℹ️ Additional command-line options are available to run other experimental settings depending on the environment configuration. See the script help (--help) for more details.

---

📄 License

This project is licensed under the Creative Commons Attribution 4.0 International (CC BY 4.0) License.