Vision-Augmented Object Detection and Pose Estimation Using ArUco Markers for Real-Time Industrial Applications

Abstract

This paper presents a vision-based system for object detection and pose estimation in industrial settings. The combination of ArUco markers and computer vision enables the system to provide accurate localization and real-time orientation estimation of objects, which applies to automated assembly lines, warehouse inventory deployment, pick-and-place industrial robots, and quality control procedures. ArUco markers serve as fiducial references, enabling reliable spatial analysis: measurement of object dimensions (height and width), centroid location, horizontal and vertical distances and orientation. These capabilities enable accurate manipulation and monitoring of industrial robots, making operations more efficient and less reliant on manual intervention. The experimental results demonstrate the system's accuracy and its responsiveness to relevant events, which can be leveraged to improve the precision, reliability, and automation of various industrial applications. A quantitative comparison of processing latency, throughput, and memory consumption against a deep learning baseline is also reported, confirming that the proposed pipeline runs in real time on CPU-only hardware. This work emphasizes that integrating computer vision methods and fiducial markers can be highly efficient for industrial automation and operational improvement.