Yao, C.; Ding, L.; Lan, Y. MOR-SLAM: A New Visual SLAM System for Indoor Dynamic Environments Based on Mask Restoration. Mathematics2023, 11, 4037.
Yao, C.; Ding, L.; Lan, Y. MOR-SLAM: A New Visual SLAM System for Indoor Dynamic Environments Based on Mask Restoration. Mathematics 2023, 11, 4037.
Yao, C.; Ding, L.; Lan, Y. MOR-SLAM: A New Visual SLAM System for Indoor Dynamic Environments Based on Mask Restoration. Mathematics2023, 11, 4037.
Yao, C.; Ding, L.; Lan, Y. MOR-SLAM: A New Visual SLAM System for Indoor Dynamic Environments Based on Mask Restoration. Mathematics 2023, 11, 4037.
Abstract
The traditional Simultaneous Localization and Mapping (SLAM) systems are based on the 1
strong static assumption, and their performance will degrade significantly due to the presence of 2
dynamic objects located in dynamic environments. To decrease the effects brought by the dynamic 3
objects, based on ORB-SLAM2 system framework, a novel dynamic semantic SLAM system called 4
MOR-SLAM is presented using mask repair method, which can accurately detect dynamic objects 5
and realize high-precision positioning and tracking of the system in dynamic indoor environments. 6
First, an instance segmentation module is added to the front end of ORB-SLAM2 to distinguish 7
dynamic and static objects in the environment and obtains a preliminary mask. Next, to overcome the 8
under-segmentation problem in instance segmentation, a new mask inpainting model is proposed to 9
ensure that the integrity of object masks, which repairs large objects and small objects in the image 10
with depth value fusion method and morphological method respectively. Then, a reliable basic matrix 11
can be obtained based on the above repaired mask. Finally, the potential dynamic feature points 12
in the environment are detected and removed through the reliable basic matrix, and the remaining 13
static feature points are input into the tracking module of the system to realize the high-precision 14
positioning and tracking in dynamic environments. The experiments on the public TUM dataset 15
show that, compared with ORB-SLAM2, the MOR-SLAM improves the absolute trajectory accuracy 16
by 95.55%. In addition, compared with DynaSLAM and DS-SLAM on the high-dynamic sequences 17
(fr3/w/rpy and fr3/w/static), the MOR-SLAM improves the absolute trajectory accuracy by 15.20% 18
and 59.71%, respectively
Computer Science and Mathematics, Artificial Intelligence and Machine Learning
Copyright:
This is an open access article distributed under the Creative Commons Attribution License which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
