Drone control via gestures using MediaPipe Hands

A project by Neurons Lab explores gesture-based drone control using MediaPipe Hands, aiming to replace traditional bulky controllers with a more intuitive method. This approach leverages gesture recognition, a natural non-verbal communication method, to control drones without physical interaction, using existing cameras on drones. The project utilizes a Ryze Tello quadcopter and a modular program structure to facilitate gesture recognition and drone control, employing a custom neural network for gesture classification. The system’s flexibility allows for quick addition and modification of gestures, making it suitable for various applications, from extreme sports to industrial environments. MediaPipe Hands offers a fast and adaptable solution, enabling real-time gesture recognition and easy model retraining. Future plans include integrating personalisation features and exploring further capabilities of MediaPipe, such as face recognition for operator authorisation, while encouraging responsible AI practices. The project aims to inspire innovation in machine learning and computer vision within UAVs, inviting contributions and ideas from the community.