LLM Task Planner for Robotics in ROS 2 Humble

---

A ROS 2 Humble framework integrating Large Language Models (LLMs) with Nav2 for high-level task planning and autonomous robot navigation.

---

Table of Contents

1. Overview
2. Features
3. Project Structure
4. Installation
5. Usage

---

Overview

This project combines the reasoning capabilities of Large Language Models with ROS 2's Nav2 navigation stack. It allows a mobile robot to:

• Plan complex tasks based on high-level natural language instructions
• Navigate autonomously in a simulated environment
• Integrate sensor data and costmaps for safe movement

It is designed to be modular and simulation-ready, making it easy to extend for research or development purposes.

---

Features

• Task Planning with LLMs: Generate sequences of robot actions from natural language instructions.
• Autonomous Navigation: Leverages ROS 2 Nav2 stack for path planning, local and global costmaps.
• Simulation Ready: Compatible with Gazebo or other ROS 2 simulation environments.
• Modular Architecture: Easy to extend with new behaviors or navigation algorithms.

---

Project Structure

LLM-Task-Planner/
├── src/                      # Source packages
│   ├── llm_nav/              # LLM-based task planning package
│   │   ├── behavior_trees/   # Behavior tree XMLs and configurations
│   │   ├── launch/           # ROS 2 launch files for nodes
│   │   ├── llm_nav/          # Node implementations
│   │   ├── maps/             # Map YAML / PGM files
│   │   ├── params/           # Parameter files
│   │   ├── rviz/             # RViz visualization configs
│   │   ├── resource/         # Other resources
│   │   └── test/             # Unit and integration tests
│   └── my_gazebo_worlds/     # Custom simulation worlds
│       ├── launch/           # Launch files for simulation
│       ├── maps/             # Map YAML / PGM files
│       ├── models/           # Simulation models
│       ├── photos/           # Screenshots and images
│       └── worlds/           # Gazebo world files
├── README.md                 # Project description and instructions
└── requirements.txt          # Python dependencies

Notes:

• src/ contains all ROS 2 packages (llm_nav, my_gazebo_worlds).
• llm_nav/: main ROS2 package that implements the task planner and integration with Nav2.
• my_gazebo_worlds/: auxiliary ROS2 package that contains simulation environments and assets.

---

Installation

1. Install ROS 2 Humble

   • Follow the official guide: ROS 2 Humble Installation

2. Clone this repository

git clone https://github.com/Dacossti/LLM-Task-Planner.git
cd LLM-Task-Planner/llm_ros2_humble

3. Install ROS dependencies

rosdep install --from-paths src --ignore-src -r -y

4. Install Python dependencies

python3 -m venv venv
source venv/bin/activate
pip install --upgrade pip
pip install -r requirements.txt

4. Build the workspace

colcon build
source install/setup.bash

Usage

• Launch LLM parsing, Navigation and Gazebo+RViz Simulation:

ros2 launch llm_nav bringup.launch.py use_nav2:=true instruction:="your_instruction"

Instruction example: Bring me coffee in the dining room