Adhan Efendi
This study presents a systematic PRISMA-guided review of 24 articles from leading databases, including MDPI, Wiley, ScienceDirect, Sage, IEEE Xplore, Chemical Science, Springer, and Elsevier, with the majority of research conducted in China. The review synthesizes advancements in collaborative robotic arms, such as UR5 and KUKA LBR iiwa, equipped with RGB-D, stereo vision, or Time-of-Flight cameras, for automated medical object handling in pharmaceutical and healthcare environments. Machine learning techniques, including deep learning, reinforcement learning, and deep reinforcement learning, implemented with YOLO and Mask R-CNN, enable object detection, grasp planning, and autonomous manipulation. Key challenges identified include lighting and occlusion sensitivity, rare or complex packaging, depth perception limitations, and environmental variability, all of which can impact detection and grasping performance. Mitigation strategies such as data augmentation, synthetic data generation, multi-view simulation, and transfer learning have proven effective in enhancing robustness and operational reliability. The findings highlight implications for improving efficiency, precision, and safety. Future research should focus on hybrid AI integration, adaptive vision systems, end-to-end learning, and digital twin validation to optimize real-world deployment in pharmaceutical settings. © The Author(s), under exclusive licence to Springer Nature Switzerland AG 2025.
Department of Mechanical Engineering Education, Universitas Negeri Surabaya, Surabaya, Indonesia
