BeSimulator: A Large Language Model Powered Text-based Behavior Simulator

A modular LLM-powered framework designed to efficiently simulate complex robotic behavior architectures, such as Behavior Trees (BTs), Finite State Machines (FSMs), and Hierarchical Task Networks (HTNs), as an effort towards behavior simulation in text-based environments.

📦 Installation

Clone the repository:

git clone https://github.com/Dawn888888/BeSimulator.git
cd BeSimulator

Create a virtual environment (recommended):

conda create --name besimulator python=3.9
conda activate besimulator

Install the required dependencies:
```
pip install -r requirements.txt
```

⚙️ Configuration

Before running BeSimulator, you must set up your API keys.

export OPENAI_API_KEY=<your_api_key>

Replace the placeholder text with your actual API keys.

🚀 Quick Start

Here is a basic example to run a Behavior Tree (BT) simulation using the DeepSeek model.

python main.py --llm_model deepseek-chat --category good --run_task_id 1

📚 Citation

If you use any of this work, it would be really nice if you could please cite 🥺 :

@article{wang2024besimulator,
  title={BeSimulator: A Large Language Model Powered Text-based Behavior Simulator},
  author={Wang, Jianan and Li, Bin and Wang, Xueying and Li, Fu and Wu, Yunlong and Chen, Juan and Yi, Xiaodong},
  journal={arXiv preprint arXiv:2409.15865},
  year={2024}
}