![](https://digilib.itb.ac.id/assets/icon/pdf.gif)
Abstrak - Lalu Muhamad Alhadad
Terbatas  Irwan Sofiyan
» Gedung UPT Perpustakaan
Terbatas  Irwan Sofiyan
» Gedung UPT Perpustakaan
The rapid advancement of unmanned aerial vehicle (UAV) technology has led to its widespread use in various sectors like agriculture, industry, and logistics for tasks such as inspection and inventory management. Drone swarming is increasingly employed in industrial applications to enhance mission efficiency. However, challenges persist in indoor settings where GPS connectivity is unavailable, necessitating innovative navigation solutions. This study aims to investigate the potential and progress of a visual-based navigation method for indoor swarming drones. It involves integrating an external camera with drone control commands to enable drones to fly collectively using a visual control system. The research assesses the effectiveness of two detection and tracking systems, YOLO with OpenCV and the CSRT algorithm, for their application in a visual-based navigation system for indoor swarming drones. The study combines coordinate data from the detection and tracking algorithms with control algorithms to guide drones through pitch and roll movements. Experimental data includes performance metrics from the detection and tracking algorithms, coordinate data for drone movements, sensor data from DJI Tello drones to validate sensor performance, and results from swarming behavior assessment and flight trajectory tests. The results indicate that while YOLO with OpenCV shows promise, it has high CPU usage and latency issues. On the other hand, the CSRT algorithm demonstrates better tracking accuracy with lower CPU utilization. Swarming behavior evaluation reveals the drones' ability to exhibit cohesion, migration, and separation despite communication delays and sensor noise. However, drones with lower-quality sensors display erratic movement, emphasizing the need for further optimization. The study highlights the significant potential of visual-based navigation for indoor swarming drones, with the CSRT algorithm proving most effective. Furthermore, system optimization and calibration are crucial to ensure reliable operation in indoor environments.