This research aimed to develop an indoor drone guidance system that relies on human hand gestures for control, as Global Positioning System (GPS) is unavailable indoors. A computer vision system was implemented using Mediapipe and a Convolutional Neural Network (CNN) to recognize and interpret hand gestures captured by a drone camera. The system was trained on a dataset of 10 distinct hand gestures, achieving a 95% accuracy rate with rapid processing time. Real-world tests were conducted using a DJI Tello drone. The developed system provides a user-friendly, accessible method for controlling drones indoors, requiring no prior experience.
Computer Vision Implementation in a Hand Gesture-Based Indoor Drone Guidance System
Springer Proceedings Phys.
International Seminar on Aerospace Science and Technology ; 2024 ; Bali, Indonesia September 17, 2024 - September 17, 2024
Proceedings of the 10th International Seminar on Aerospace Science and Technology; ISAST 2024; 17 September, Bali, Indonesia ; Chapter : 36 ; 333-343
2025-02-15
11 pages
Article/Chapter (Book)
Electronic Resource
English
Human-Drone Interaction , Computer Vision , Gesture Recognition , Mediapipe , Hybrid Model , UAV Physics , Space Sciences (including Extraterrestrial Physics, Space Exploration and Astronautics) , Aerospace Technology and Astronautics , Climate, general , Ecology , Astronomy, Astrophysics and Cosmology , Measurement Science and Instrumentation , Physics and Astronomy