Development and Validation of a Low-Cost Vision-Based Indoor Motion Capture System for GPS-Free UAV Navigation

An accurate and reliable motion capture (MoCap) system is essential for unmanned aerial vehicle (UAV) development, particularly for indoor flight tests.Existing wireless positioning methods, such as ultra-wideband, Wi-fi, or Bluetooth, are appealing because of their relatively low hardware requirements and ease of integration, but they are limited by multipath interference and inaccuracy.Computer vision-based systems offer higher accuracy; however, commercial solutions such as Vicon or Optitrack are costly and depend on proprietary software.This paper presents an accurate, vision-based, low-cost MoCap system for indoor UAV tracking.The system combines multiple Raspberry Pi cameras with hardware and software synchronization, reflective markers as tracking points, and Robot Operating System (ROS) middleware for communication.Experimental validation compares position data from the MoCap system with estimates from the PX4 flight controller.Results demonstrate good tracking performance across various UAV flight trajectories, with root mean square errors in 3D consistently below 0.06 m and no directional deviations exceeding 0.05 m.