GPT Voice-Driven Natural Language Control for ArduPilot-Guided UAVs
Mohammad Al Khawaldah, Omar Rawhi Daoud, Saqer S. Alja’afreh, Aser M. Matarneh · 2025
This research work introduces a new system that utilizes Generative Pretrained Transformer (GPT) Large Language Models (LLMs) to enable language-based control for Unmanned Aerial Vehicles (UAVs). The system enables nonexpert users to easily control UAVs using natural language. It integrates GPT-4o-mini model with the widely used open-source ArduPilot Software. The developed system supports both audio and written inputs and accommodates multiple languages. The LLM reasons about the request and generates the required functionality. OpenAI's API (Application Programming Interface) and Function Calling are used to implement the task.