Experimental Analysis of Pose Estimation Based on ArUco Markers

Pose estimation is a critical task in various domains that requires motion capture systems. In recent years, camera-based motion capture technologies have grown due to advances in artificial vision. Camera-based motion capture systems provide accurate results in pose estimation at a significant cost and demand high computational resources. Fiducial markers, like ArUco, have emerged as cost-effective alternatives, leveraging simplicity, low computational cost, and adaptability. However, there are few experimental analyses about the accuracy of an ArUco-based motion capture system. This study proposes an experimental performance analysis of an ArUco-based motion capture system with respect to a commercial system manufactured by Optitrack. The ArUco-based motion capture system is developed by means of an EPS webcam and a ZED stereo camera. Through a series of experiments, the pose estimation error is evaluated in static and dynamic scenarios in different regions of the visible camera area. Furthermore, the relationship between pose estimation error and the utilisation of a bundle of ArUco markers is analysed. The results show that the proposed ArUco-based motion capture system provides moderate accuracy.