Fly a Drone Using an LLM with ROS 2 + Gazebo

About this course

Learn how to build and control an AI-powered simulated drone using Large Language Models, ROS 2, PX4, Gazebo, Ollama, and YOLO.You will be able to give instructions such as:

• “Take off and hover at 5 meters.”
• “Fly in a heart shape.”
• “Search for a person, then approach the person.”

This beginner-friendly course takes you from a clean Ubuntu 24.04 installation to a complete drone system that can understand natural-language instructions, execute autonomous flight behaviors, and react to objects detected through its camera.

You will begin by setting up ROS 2 Jazzy, Gazebo, and PX4 Autopilot, then launch and test a simulated X500 drone. You will also learn how PX4 communicates with ROS 2 through the Micro XRCE-DDS Agent and how the simulated camera is bridged from Gazebo into ROS 2.

Next, you will install Ollama and run a local Large Language Model that converts simple English instructions into drone actions. Instead of writing low-level commands.

You will also integrate YOLO object detection with the drone camera, allowing the system to detect objects such as people and vehicles and use those detections during autonomous missions.

Finally, you will launch the complete LLM-controlled drone system and experiment with flight patterns, object search and approach behaviors, multi-step missions, camera visualization, and drone path visualization using RViz.

No previous experience with ROS 2, PX4, Gazebo, or LLM-based robotics is required. The course is designed to give beginners a practical robotics and AI project they can extend, demonstrate in a portfolio, or discuss during robotics and AI job interviews.

Prices, discounts and availability are set by Udemy. We may earn a commission when you purchase through links on this site.