YOLO-Based 3D Perception for UVMS Grasping

• Published in: Journal of marine science and engineering Vol. 12; no. 7; p. 1110
• Main Authors: Chen, Yanhu, Zhao, Fuqiang, Ling, Yucheng, Zhang, Suohang
• Format: Journal Article
• Language: English
• Published: Basel MDPI AG 01.07.2024
• Subjects: Ablation; Accuracy; Algorithms; deep learning; Depth perception; Distance measurement; Efficiency; Grasping; Grasping (robotics); Localization; marine organism grasping; Measurement techniques; Object recognition; Optimization; Robotics; Space perception; Target recognition; underwater binocular ranging; underwater object detection; Underwater robots; Underwater vehicles
• ISSN: 2077-1312; 2077-1312
• DOI: 10.3390/jmse12071110

This study develops a YOLO (You Only Look Once)-based 3D perception algorithm for UVMS (Underwater Vehicle-Manipulator Systems) for precise object detection and localization, crucial for enhanced grasping tasks. The object detection algorithm, YOLOv5s-CS, integrates an enhanced YOLOv5s model with C3SE attention and SPPFCSPC feature fusion, optimized for precise detection and two-dimensional localization in underwater environments with sparse features. Distance measurement is further improved by refining the SGBM (Semi-Global Block Matching) algorithm with Census transform and subpixel interpolation. Ablation studies highlight the YOLOv5s-CS model’s enhanced performance, with a 3.5% increase in mAP and a 6.4% rise in F1 score over the base YOLOv5s, and a 2.1% mAP improvement with 15% faster execution than YOLOv8s. Implemented on a UVMS, the algorithm successfully conducted pool grasping experiments, proving its applicability for autonomous underwater robotics.