Comparative Performance of YOLO Versions for Pallet Detection SCARA Robot
DOI:
https://doi.org/10.24036/05dbyr50Keywords:
YOLO, Comparative Study, Pallet Detection, Machine LearningAbstract
The rapid evolution of the You Only Look Once (YOLO) object detection family from YOLOv8 to YOLO26 has produced architecturally diverse models whose relative suitability for structured industrial detection tasks remains unevaluated. This study aims to conduct a controlled comparative performance evaluation of six YOLO variants (YOLOv8, YOLOv9, YOLOv10, YOLO11, YOLOv12, and YOLO26) on an industrial pallet detection dataset to identify the optimal model for Vision-Guided Robotics (VGR) pick and place automation. All models were trained under identical conditions 50 epochs, batch size 16, 640×640 input resolution on a dataset of 4,366 annotated images across three classes (pallet_pick, pallet_place_1, pallet_place_2) using a 70/20/10 train/validation/test split. Performance was assessed using Precision, Recall, mAP@0.5, mAP@0.5:0.95, and training time. All six models achieved perfect Precision (1.0000), Recall (1.0000), and mAP@0.5 (0.9950). Differentiation emerged at mAP@0.5:0.95, where YOLOv9m achieved the highest final-epoch value (0.9776) and YOLO26m achieved the highest peak value (0.9805) with the fastest training time (80.8 minutes, 26.3% faster than YOLOv8). YOLO26 is recommended as the optimal model for industrial pallet detection, offering the best combination of localization accuracy, training efficiency, and deployment readiness through its NMS-free and DFL-free architecture.
Downloads
References
[1] C. Gabellieri, A. Palleschi, L. Pallottino and M. Garabini, "Autonomous Unwrapping of General Pallets: A Novel Robot for Logistics Exploiting Contact-Based Planning," IEEE Transactions on Automation Science and Engineering, vol. 20, no. 2, pp. 1194-1211, April 2023.
[2] C. D. Santina, “Learning from humans how to grasp: A data-driven architecture for autonomous grasping with anthropomorphic soft hands,” IEEE Robot. Autom. Lett., vol. 4, no. 2, pp. 1533–1540, Apr. 2019.
[3] A. Singh, V. Kalaichelvi, and R. Karthikeyan, "A survey on vision guided robotic systems with intelligent control strategies for autonomous tasks," Cogent Eng., vol. 9, no. 1, p. 2050020, 2022, doi: 10.1080/23311916.2022.2050020.
[4] J. -S. Kang, J. K. Kang, S. Son, S. H. Park, E. -J. Jung and G. R. Cho, "Image Enhancement Method for Detection in Logistics Loading and Unloading Robots," IEEE Access, vol. 13, pp. 143678-143687, 2025, doi: 10.1109/ACCESS.2024.3434599.
[5] R. R. Arakerimath, A. H. Badve, D. P. Deshmukh, M. N. Chandan and A. Badadhe, "Performance Modelling and Simulation of Cobot for Pick and Place Application Using RoboDK," 2025 International Conference on Emerging Technologies in Electronics and Green Energy (ICETEG), MYSORE, India, 2025, pp. 1-6, doi: 10.1109/ICETEG66194.2025.11473126.
[7] R. Sapkota et al., "YOLO advances to its genesis: a decadal and comprehensive review of the You Only Look Once (YOLO) series," Artif. Intell. Rev., vol. 58, no. 9, 2025, doi: 10.1007/s10462-025-11253-3.
[8] S. Kang, Z. Hu, L. Liu, K. Zhang, and Z. Cao, "Object Detection YOLO Algorithms and Their Industrial Applications: Overview and Comparative Analysis," Electronics, vol. 14, no. 6, pp. 1–36, 2025, doi: 10.3390/electronics14061104.
[9] R. Sapkota and M. Karkee, "Ultralytics YOLO Evolution: An Overview of YOLO26, YOLO11, YOLOv8, and YOLOv5 Object Detectors for Computer Vision and Pattern Recognition," arXiv preprint, arXiv:2510.09653, 2025.
[10] G. Jocher et al., "Ultralytics YOLOv8," Ultralytics, 2023. [Online]. Available: https://github.com/ultralytics/ultralytics
[11] C.-Y. Wang, I-H. Yeh, and H.-Y. M. Liao, "YOLOv9: Learning What You Want to Learn Using Programmable Gradient Information," arXiv preprint, arXiv:2402.13616, 2024.
[12] A. Wang et al., "YOLOv10: Real-Time End-to-End Object Detection," Adv. Neural Inf. Process. Syst., vol. 37, pp. 107984–108011, 2024.
[13] X. Ren, Z. Liu, H. Cheng and L. Chen, "DDR-YOLO: An Efficient Detection Model Based on YOLO11s for Steel Surface Defect," in IEEE Access, vol. 13, pp. 147202-147217, 2025, doi: 10.1109/ACCESS.2025.3599978
[14] Y. Tian, Q. Ye, and D. Doermann, "YOLOv12: Attention-Centric Real-Time Object Detectors," arXiv preprint, arXiv:2502.12524, 2025.
[15] R. Sapkota, R. H. Cheppally, A. Sharda, and M. Karkee, "YOLO26: Key Architectural Enhancements and Performance Benchmarking for Real-Time Object Detection," arXiv preprint, arXiv:2509.25164, 2025.
[16] Y. Gu, "CRP-YOLO26: A Lightweight Improved YOLO26 Algorithm for Small Object Detection in UAV Images," 2026 5th International Conference on Artificial Intelligence, Internet of Things and Cloud Computing Technology (AIoTC), Nanchang, China, 2026, pp. 683-689, doi: 10.1109/AIoTC71250.2026.11689104.
[17] R. Sapkota et al., "Comprehensive performance evaluation of YOLOv12, YOLO11, YOLOv10, YOLOv9 and YOLOv8 on detecting and counting fruitlet in complex orchard environments," Smart Agric. Technol., 2026, doi: 10.1016/j.atech.2026.
[18] P. Nizieniec, M. Iwanowski et al., "YOLO Object Detectors for Robotics a Comparative Study," arXiv preprint, arXiv:2603.27029, 2026.
[19] A. Minsandi et al., “Design and Implementation of Robot Abu Robocon Using Joystik Wireless Based on Extrasenory Percepsion,” J. Ind. Autom. Electr. Eng., vol. 01, no. 01, pp. 39–45, 2024..
[20] T. P. Kapusi, T. I. Erdei, G. Husi, and A. Hajdu, "Application of Deep Learning in the Deployment of an Industrial SCARA Machine for Real-Time Object Detection," Robotics, vol. 11, no. 4, 2022, doi: 10.3390/robotics11040069.
[21] V. A. Putra, “Cycle time analysis of automation system using PLC and HMI based Scara Robot,” J. Ind. Autom. Electr. Eng., vol. 02, no. 01, pp. 112–118, 2025.
[22] P. Nizieniec and M. Iwanowski, "YOLO Object Detectors for Robotics a Comparative Study," arXiv preprint, arXiv:2603.27029, 2026.
