针对水下机器人对接过程中近距离引导定位的问题，为实现准确可靠的水下对接，设计了一种基于神经网络的水下对接引导算法。算法主要分为两部分：1）检测部分，该检测算法区别于传统的图像处理，基于YOLOv3的神经网络，加入DenseNet思想，实现了一种鲁棒性强，可靠以及准确率高的目标检测算法。区分了目标物与无关背景信息，用于提供水下对接站在二维图像下的位置信息。2）位姿估计部分，用于恢复水下机器人与中继器之间的相对位置信息，从而能够根据相对位置信息，完成近距离的水下对接引导。经过实验对比分析，验证了算法的可行性，并证明了改进的目标检测算法能够更精确地完成对中继器的位置检测，提高了对位姿估计的准确率。
Aiming at the problem of short-range guidance and positioning in the docking process of underwater robots, in order to achieve accurate and reliable underwater docking, this paper designs an underwater docking guidance algorithm based on neural network. The algorithm consists of two parts: 1. The detection module, which is different from traditional image processing, is based on the yolov3 neural network, and joins the DenseNet idea to achieve a robust, reliable and accurate target detection algorithm. The algorithm is used to provide the location information of the underwater docking station and obtain the docking station information under the two-dimensional image. 2. The pose estimation module is used to restore the relative position information between the underwater robot and the docking station, so that the underwater docking task can be completed according to the relative position information. Experimental comparison and analysis have verified the feasibility of the algorithm and proved that the improved neural network detection algorithm can be more effectively applied to the position detection of the docking station, which improves the accuracy of the pose estimation.
2021,43(9): 102-107 收稿日期：2020-09-10
DOI：10.3404/j.issn.1672-7649.2021.09.020
分类号：TP242.3
作者简介：冯晓晨(1995-)，男，硕士研究生，研究方向为水下机器人和计算机视觉
参考文献：
[1] BELLINGHAM J. G. (2016) Autonomous underwater vehicle docking. In: Dhanak, M. R. and Xiros, N. I. (eds) Springer International Publishing, 387–406.
[2] LI Y, JIANG Y, CAO J, et al. Auv docking experiments based on vision positioning using two cameras[J]. Ocean Engineering, 2015, 110: 163–173
[3] MAKI T, SHIROKU R, SATO Y, et al. (2013) Docking method for hovering type auvs by acoustic and visual positioning[C]// In: 2013 IEEE International Underwater Technology Symposium(UT), Tokyo, Japan, 5–8 March 2013, 1–6.
[4] H. SINGH, G. JAMES Docking for an autonomous ocean sampling network[J]. IEEE Journal of Oceanic Engineering, 2001: 498–514
[5] ZENG A, YU Kuanting, SONG Shuran, et al. Multi-view self-supervised deep learning for 6d pose estimation in the amazon picking challenge[C]// IEEE International Conference on Robotics and Automation. Piscataway, NJ: IEEE Press, 2017: 1386−1383.
[6] GIRSHICK R, DONAHUE J, DARRELL T, et al. Rich feature hierarchies for accurate object detection and semantic segmentation[C]//IEEE Conference on Computer Vision & Pattern Recognition. IEEE Computer Society, 2014: 580−587.
[7] GIRSHICK R B. Fast R-CNN[C]//The IEEE International Conference on Computer Vision (ICCV), 2015: 1440–1448.
[8] DAI J, LI Y, HE K, et al. R-FCN: object detection via region-based fully convolutional networks[C]//Conference and Workshop on Neural Information Processing Systems, 2016: 379–387.
[9] REDMON J, FARHADI A. YOLO9000: better, faster, stronger[C]//The IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2017: 7263−7271.
[10] LIU W, ANGUELOV D, ERHAN D, et al. SSD: single shot multibox detector[C]//European Conference on Computer Vision. 2016: 21–37.
[11] HE K, ZHANG X, REN S, et al. Deep residual learning for image recognition[C]//IEEE Conference on Computer Vision and Pattern Recognition(CVPR). IEEE Computer Society, 2016: 770−778.
[12] HUANG G, LIU Z, LAURENS V D M. Densely connected convolutional networks[C]//IEEE conference on Computer Vision and Pattern Recognition, 2017: 2261−2269.
[13] 王丙乾, 唐元贵. 面向ARV水下对接的视觉引导信息特征点匹配算法研究[J]. 海洋技术学报, 2018, 37(5): 9–15
[14] ZHANG Z. Flexible camera calibration by viewing a plane from unknown orientations[C]// Proceedings of the Seventh IEEE International Conference on Computer Vision. IEEE, 1999.
[15] ZHANG Z. A flexible new technique for camera calibration[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2000, 22(11): 1330–1334
[16] 伊文康, 高宏伟. 基于视觉的水下机器人自主对接引导算法研究[D]. 沈阳: 沈阳理工大学, 2017.
