Real-Time Hand Gesture Controlled UAV Using Computer Vision Techniques.

Abstract

Unmanned Aerial Vehicles (UAVs) are widely deployed in applications such as surveillance, inspection, and environmental monitoring. Conventional UAV control methods depend on handheld controllers or mobile devices, which may be impractical in situations requiring hands-free or rapid response interaction. This study proposes a real-time, vision-driven hand gesture control system that enables intuitive and touchless UAV operation. The system captures live video using a camera and processes it through a computer vision pipeline to detect and track hand movements. Using landmark-based hand modelling , key features are extracted and analyzed to identify predefined gestures based on spatial relationships among hand joints. The implementation leverages Python along with OpenCV and MediaPipe Hands for efficient gesture recognition. Recognized gestures are mapped to specific flight commands and transmitted to a DJI Tello drone via its Software Development Kit (SDK), allowing real-time execution of actions such as takeoff, landing, directional navigation, and altitude adjustment. To enhance operational safety, the system integrates mechanisms such as gesture validation delays, emergency stop controls, and automatic landing triggered by low battery levels or communication disruptions.Experimental evaluation demonstrates that the system achieves reliable gesture recognition with minimal latency, resulting in smooth and stable drone maneuvering. The proposed framework eliminates the need for traditional input devices and offers a practical solution for natural human–drone interaction. This approach is particularly beneficial for applications in search and rescue, surveillance, and advanced human–machine interfaces, demonstrating its potential as a scalable and effective UAV control paradigm.