ROSA Agent for TurtleBot3 Navigation (ROS 1 Noetic)

This repository integrates the ROSA agent framework with a simulated TurtleBot3 robot using ROS 1 Noetic and the standard ROS Navigation stack (move_base).

Overview

The setup uses two Catkin workspaces to manage separate Python environments:

• turtlebot3_ws: For TurtleBot3 simulation and navigation (Python 3.8).
• zrosa_ws: For ROSA agent code using a Python 3.9 virtual environment.

The ROSA agent accepts natural language commands and sends goals to the /move_base action server to control TurtleBot3.

Project Structure

final-project/
├── turtlebot3_ws/               # Workspace for TurtleBot3 simulation/navigation
│   └── src/
│       ├── turtlebot3/
│       ├── turtlebot3_msgs/
│       └── turtlebot3_simulations/
├── zrosa_ws/                    # Workspace for ROSA Agent (Python 3.9 venv)
│   ├── venv_rosa/               # Python 3.9 virtual environment
│   └── src/
│       └── llm-ur-control/
│           └── ur_agent/       # ROSA agent package
│               ├── scripts/
│               │   ├── ur_agent.py
│               │   ├── locations.py
│               │   ├── llm.py
│               │   ├── prompts.py
│               │   └── help.py
│               ├── package.xml
│               └── CMakeLists.txt
└── README.md                   # This file

Prerequisites

1. Operating System: Ubuntu 20.04 LTS
2. ROS Version: ROS 1 Noetic
3. Python:
   • System: Python 3.8
   • Additional: Python 3.9 (python3.9, python3.9-dev, python3.9-venv)
4. Dependencies:

   sudo apt install git python3.9 python3.9-dev python3.9-venv \
       ros-noetic-joy ros-noetic-teleop-twist-joy \
       ros-noetic-teleop-twist-keyboard ros-noetic-slam-gmapping \
       ros-noetic-navigation ros-noetic-amcl \
       ros-noetic-map-server ros-noetic-move-base \
       ros-noetic-turtlebot3-simulations

5. Map Files: Create via:

   rosrun map_server map_saver -f ~/map

6. OpenAI API Key: Required for LLM interaction.

Setup Instructions

Step 1: Clone the Repository

git clone https://github.com/ASIFXS/final-project.git
cd final-project

Step 2: Clean ~/.bashrc

• Ensure only this line is sourced:

  source /opt/ros/noetic/setup.bash

• Comment or remove any lines that source workspace-specific setup.bash files.

Step 3: Install Python 3.9

sudo add-apt-repository ppa:deadsnakes/ppa
sudo apt update
sudo apt install python3.9 python3.9-dev python3.9-venv

Step 4: Build turtlebot3_ws

cd ~/final-project/turtlebot3_ws
source /opt/ros/noetic/setup.bash
rosdep install --from-paths src --ignore-src -r -y
catkin_make

Step 5: Setup zrosa_ws

cd ~/final-project/zrosa_ws
python3.9 -m venv venv_rosa
source venv_rosa/bin/activate
pip install --upgrade pip
pip install python-dotenv langchain openai rich pyinputplus numpy
source /opt/ros/noetic/setup.bash
catkin_make

Step 6: Configure Locations

Edit locations.py:

gedit ~/final-project/zrosa_ws/src/llm-ur-control/ur_agent/scripts/locations.py

Modify the LOCATIONS dictionary with appropriate (x, y, yaw) values.

Step 7: Add OpenAI API Key

cd ~/final-project/zrosa_ws
gedit .env

Add:

OPENAI_API_KEY='sk-YourActualKeyHere'

Running Instructions

Terminal Setup

Open 3 terminals. In each, export the model:

export TURTLEBOT3_MODEL=burger

Terminal 1: Launch Gazebo

cd ~/final-project/turtlebot3_ws
source devel/setup.bash
roslaunch turtlebot3_gazebo turtlebot3_world.launch

Terminal 2: Launch Navigation

cd ~/final-project/turtlebot3_ws
source devel/setup.bash
roslaunch turtlebot3_navigation turtlebot3_navigation.launch map_file:=$HOME/map.yaml

Use the 2D Pose Estimate tool in RViz to localize the robot.

Terminal 3: Launch ROSA Agent

cd ~/final-project/zrosa_ws
source venv_rosa/bin/activate
source devel/setup.bash
rosrun ur_agent ur_agent.py

Interact with commands like:

> go to L1
> navigate to CENTER

Customization

• Location Points: Modify locations.py
• API Key: Ensure .env file is correctly set

Troubleshooting

• catkin_make errors: Ensure correct dependencies in package.xml and CMakeLists.txt. Source /opt/ros/noetic/setup.bash first.
• ModuleNotFoundError: Activate venv and install missing packages.
• move_base not found: Ensure Navigation stack is running.
• Poor localization: Re-estimate pose in RViz.
• Navigation fails: Check map for obstacles or localization accuracy.

---

Enjoy controlling your TurtleBot3 using natural language with ROSA!