|本期目录/Table of Contents|

[1]潘海鹏,刘培敏,马淼.基于语义信息与动态特征点剔除的SLAM算法[J].浙江理工大学学报,2022,47-48(自科五):764-773.
 PAN Haipeng,LIU Peimin,MA Miao.SLAM algorithm based on semantic information and  the elimination of dynamic feature points[J].Journal of Zhejiang Sci-Tech University,2022,47-48(自科五):764-773.
点击复制

基于语义信息与动态特征点剔除的SLAM算法()
分享到:

浙江理工大学学报[ISSN:1673-3851/CN:33-1338/TS]

卷:
第47-48卷
期数:
2022年自科第五期
页码:
764-773
栏目:
出版日期:
2022-09-10

文章信息/Info

Title:
SLAM algorithm based on semantic information and  the elimination of dynamic feature points
文章编号:
1673-3851 (2022) 09-0764-10
作者:
潘海鹏刘培敏马淼
浙江理工大学 机械与自动控制学院,杭州 310018
Author(s):
PAN Haipeng LIU Peimin MA Miao
School of Mechanical Engineering and Automation, Zhejiang  Sci-Tech University, Hangzhou 310018, China
关键词:
同时定位与地图构建动态环境动态特征点剔除注意力机制损失函数
分类号:
TP391
文献标志码:
A
摘要:
传统的同时定位与地图构建(Simultaneous localization and mapping,SLAM)算法在现实场景中易受动态物体及背景的影响,针对该问题提出了一种将语义分割与动态特征点剔除相结合的动态SLAM算法,以实现动态场景地图的构建。首先,根据多层通道注意力和空间注意力机制,构造特征融合网络MulAttenNet(Multilayer attention network),并进行语义分割,剔除场景中运动概率大的物体,粗略估计相机位姿;其次,根据相机位姿和深度信息剔除动态区域;最后,利用剔除后的特征点进行地图的构建。对MulAttenNet网络和动态SLAM算法进行实验,以验证算法的有效性,实验结果表明:该算法构造的MulAttenNet网络能有效提高语义分割的准确性,平均像素准确度提高4.05%,均交并比提高2.60%;将该算法构建的动态SLAM算法与现有SLAM算法相比,建图的绝对位姿误差和相对位姿误差都有所缩小。该算法能在动态场景下构建高精度的语义地图。

参考文献/References:

[1]Davison A J, Reid I D, Molton N D, et al. MonoSLAM: Realtime single camera SLAM[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2007, 29(6): 1052-1067.

[2]Engel J, Schps T, Cremers D. LSDSLAM: Largescale direct monocular SLAM[C]//European Conference on Computer Vision(ECCV 2014). Cham: Springer, 2014: 834-849.

[3]MurArtal R, Tards J D. ORBSLAM2: An opensource SLAM system for monocular, stereo, and RGBD cameras[J]. IEEE Transactions on Robotics, 2017, 33(5): 1255-1262.

[4]Newcombe R A, Fox D, Seitz S M. Dynamicfusion: Reconstruction and tracking of nonrigid scenes in realtime[C]//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition. Boston, MA, USA: IEEE, 2015: 343-352.

[5]Bescos B, Fcil J M, Civera J, et al. DynaSLAM: Tracking, mapping, and inpainting in dynamic scenes[J]. IEEE Robotics and Automation Letters, 2018, 3(4): 4076-4083.

[6]Chen X, Milioto A, Palazzolo E, et al. SuMa++: efficient LiDARbased semantic SLAM[C]//2019 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS). Macau, China: IEEE, 2019: 4530-4537.

[7]Zhang J, Henein M, Mahony R, et al. VDOSLAM: A visual dynamic objectaware SLAM system. (2020-5-22)[2022-03-24].

[8]Wimbauer F, Yang N, Von Stumberg L, et al. MonoRec: Semisupervised dense reconstruction in dynamic environments from a single moving camera[C]//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. Nashville, TN, USA: IEEE, 2021: 6108-6118.

[9]Vaswani A, Shazeer N, Parmar N, et al. Attention is all you need[C]//Advances in Neural Information Processing Systems. Long Beach, CA, USA: NIPS, 2017: 5998-6008.

[10]邓远远, 沈炜. 基于注意力反馈机制的深度图像标注模型[J]. 浙江理工大学学报(自然科学版), 2019, 41(2): 208-216.

undefined

备注/Memo

备注/Memo:
收稿日期: 2022-03-24
网络出版日期:2022-06-02
基金项目: 浙江省自然科学基金项目(LQ19F030014)
作者简介: 潘海鹏(1965-),男,河南濮阳人,教授,硕士,主要从事智能信息处理方面的研究
通信作者: 马淼,E-mail:mamiao@zstu.edu.cn
更新日期/Last Update: 2022-09-07