作者：张青春, 王旺, 杨广栋

作者单位：淮阴工学院自动化学院，江苏 淮安 223003

关键词：立体视觉;机械臂;跨尺度;三维重建

摘要：

为提高工业抓取物料的准确性与速度，针对质量为70 kg物料抓取困难的问题，以工业摄像头、计算机、光源和工业机械手为硬件基础，5G移动通信为系统与上位机之间的数据交互，构建多目立体视觉的机械臂智能控制系统。提出一种改进的跨尺度引导滤波算法，在代价聚合阶段利用奇数采样层使用基于高斯拉普拉斯算子的引导图滤波、偶数采样层使用引导滤波的方法对不同高斯采样层的代价卷进行代价聚合，并在视差精细化阶段加入亚像素增强的方法。最后在Matlab和OpenCV平台进行立体匹配实验和机械臂运动轨迹的仿真。实验结果表明，该算法能更加准确获取物体的空间坐标，完成三维重建，机械臂能够成功完成智能抓取。

Design of industrial robot manipulator intelligent control system based on multi-view stereo vision
ZHANG Qingchun, WANG Wang, YANG Guangdong
Automation Faculty, Huaiyin Institute of Technology, Huai’an 223003, China
Abstract: To improve the accuracy and speed of industrial grabbing material. To solve the problem of difficult material grabbing with a weight of 70 kg, an intelligent control system of mechanical arm with multi-view stereo vision is proposed. The system based on industrial camera, computer, light source and industrial mechanical arm and 5G mobile communication as data interaction between the system and upper computer. A stereo matching method based on the guided filtering cross-scale transform was improved. Different cost aggregation algorithm is used for the cost aggregation of different downsampling layers. In the odd downsampling layer, guided image filtering based on Laplacian of Gaussion operator is used. At the same time, by adding the Sub-pixel enhancement in the disparity refinement stage. Finally, a jiont simulation by using Matlab and OpenCV was conducted, by the stereo matched and the motion track of the mechanical arm.The experimental results show that the algorithm can obtain the spatial coordinates of the object more accurately and complete the 3D reconstruction,the mechanical arm can successfully completed intelligent grasp.
Keywords: stereo vision;robot manipulator;cross scale;3D reconstruction
2020, 46(12):79-85  收稿日期: 2020-05-29;收到修改稿日期: 2020-06-23
基金项目:
作者简介: 张青春(1964-),男,江苏淮安市人,教授,研究方向为智能检测技术、无线传感器、机器人与物联网、虚拟仪器技术等
参考文献
[1] YANG S, MACLACH L, RIVIERE C N. Manipulator design and operation for six-degree-of-freedom hand-held tremor canceling microsurgical instrument[J]. IEEE/ASME Transactions on Mechatronics, 2015, 20(2): 761-772
[2] DOGANGIL G, DAVIES B L, RODRIGUEZY B F. A Review of medical robotics for minimally invasive soft tissue surgery[J]. Proceedings of the Institution of Mechanical Engineers Part H Journal of Engineering in Medicine, 2010, 224(5): 653
[3] 肖振楠, 申燚, 倪辰旖. 六自由度机械手的运动轨迹规划与仿真[J]. 机床与液压, 2018, 46(15): 58-63
[4] 宋现义. 基于机器视觉的机器人智能搬运系统研究[D]. 哈尔滨：哈尔滨理工大学, 2018.
[5] 钱志敏. 基于多目视觉的工业机器人应用研究[D]. 成都：电子科技大学, 2019.
[6] ZHANG K, FANG Y, MIN D, et al. Cross-scale cost aggregation for stereo matching[C]//2014 IEEE Conference on Computer Vision and Pattern Recognition, Columbus, OH, USA. USA: IEEE, 2014: 1590-1597. DOI: 10.1109/CVPR.2014.206.
[7] TANG C M, JIANG A. Improved stereo matching algorithm based on cross-scale cost aggregation[J]. Computer Engineering, 2016, 42(11): 272-276, 280
[8] 姚莉, 刘助奎, 王秉凤. 基于金字塔变换跨尺度代价聚合的立体匹配[J]. 系统仿真学报, 2016, 28(9): 2227-2234
[9] 周博, 秦岭, 龚伟. 基于高斯拉普拉斯算子的加权引导图滤波立体匹配算法[J]. 激光与光电子学进展, 2019, 56(10): 234-240
[10] 高申勇. 高精度局部立体视觉匹配法研究[D]. 杭州：浙江大学, 2018.
[11] ZHANG Q, XU L, JIA J Y. 100 + Times faster weighted median filter (WMF)[C]//Proceedings of Computer Vision and Pattern Recognition. IEEE, 2014: 2830-2837.
[12] YOON K J, KWEON I S. Adaptive support-weight approach for correspondence search[J]. IEEE Transactions on Pattern Analysis & Machine Intelligence (S0162-8828), 2006, 28(4): 650-656