Instance recognition and semantic mapping based on visual SLAM

被引：0

作者：

Wu H. ^{[1
]}

Chi J. ^{[1
]}

Tian G. ^{[1
]}

机构：

[1] School of Control Science and Engineering, Shandong University, Jinan

来源：

Huazhong Keji Daxue Xuebao (Ziran Kexue Ban)/Journal of Huazhong University of Science and Technology (Natural Science Edition) | 2019年 / 47卷 / 09期

关键词：

3D semantic mapping; Instance recognition; Instance segmentation; Object detection; Object tracking; SLAM (simultaneous localization and mapping);

D O I：

10.13245/j.hust.190909

中图分类号：

学科分类号：

摘要：

The majority of research about semantic SLAM (simultaneous localization and mapping) to date require a priori known 3D (three dimensional) object models or revolve around mapping with a few object categories but neglect separating object individual.In view of these problems,an instance recognition and 3D semantic mapping method based on visual SLAM was proposed by combining the deep-learning-based instance segmentation and visual SLAM algorithm,which helps robots not only gain the navigation-oriented geometric information of environment,but also grasp the individual-oriented attribute and location of objects.The geometrical consistency constraint of image frames was used to promote the object matching and recognition results in continuous image frames and the object recognition results were used to help with semantic mapping.Finally,the instance recognition and semantic mapping system was implemented,and the simulation experiment was carried out on ICL-NUIM dataset.The experimental results show that the system can basically recognize all kinds of objects and generate the 3D semantic map of environment,which verifies the effectiveness of the method. © 2019, Editorial Board of Journal of Huazhong University of Science and Technology. All right reserved.

引用

页码：48 / 54

页数：6

共 16 条

[1]

Bao S.Y., Bagra M., Chao Y.W., Et al., Semantic structure from motion with points, regions, and objects, Proc of CVPR 2012, pp. 2703-2710, (2012)

[2]

Zhao C., Sun L., Purkait P., Et al., Dense RGB-D semantic mapping with Pixel-Voxel neural network, Sensors, 18, 9, (2018)

[3]

Whelan T., Salas-Moreno R.F., Glocker B., Et al., ElasticFusion: real-time dense SLAM and light source estimation, The International Journal of Robotics Research, 35, 14, pp. 1697-1716, (2016)

[4]

McCormac J., Handa A., Davison A., Et al., SemanticFusion: dense 3D semantic mapping with convolutional neural networks, Proc of IEEE International Con- Ference on Robotics and Automation, pp. 4628-4635, (2017)

[5]

Galindo C., Saffiotti A., Coradeschi S., Et al., Multi-hierarchical semantic maps for mobile robot ics, Proc of IROS 2005, pp. 2278-2283, (2005)

[6]

Civera J., Galvez-Lopez D., Riazuelo L., Et al., Towards semantic SLAM using a monocular camera, Proc of IROS 2011, pp. 1277-1284, (2011)

[7]

Galvez-Lopez D., Salas M., Tardos J.D., Et al., Real-time monocular object SLAM, Robotics & Autonomous Systems, 75, pp. 435-449, (2016)

[8]

Salas-Moreno R.F., Newcombe R.A., Strasdat H., Et al., SLAM++: simultaneous localisation and mapping at the level of objects, Proc of CVPR 2013, pp. 1352-1359, (2013)

[9]

He K., Gkioxari G., Dollar P., Et al., Mask R-CNN, Proc of ICCV 2017, pp. 2980-2988, (2017)

[10]

Mur-Artal R., Tardos J.D., ORB-SLAM2: an open-source SLAM system for monocular, stereo, and rgb-d cameras, IEEE Transactions on Robotics, 33, 5, pp. 1255-1262, (2017)

← 1 2 →