An Approach of Dynamic Object Removing for Indoor Mapping Based on UGV SLAM

Tong Yu, Jingbin Liu, Yuwei Chen, Jian TangSHILAP 2015

Abstract

The study of indoor mapping for Location Based Service (LBS) becomes more and more popular in recent years. LiDAR SLAM based mapping method seems to be a promising indoor mapping solution. However, there are some dynamic objects such as pedestrians, indoor vehicles, etc. existing in the raw LiDAR range data. They have to be removal for mapping purpose. In this paper, a new approach of dynamic object removing called Likelihood Grid Voting (LGV) is presented. It is a model free method and takes full advantage of the high scanning rate of LiDAR, which is moving at a relative low speed in indoor environment. In this method, a counting grid is allocated for recording the occupation of map position by laser scans. The lower counter value of this position can be recognized as dynamic objects and the point cloud will be removed from map. This work is a part of algorithms in our self- developed Unmanned Ground Vehicles (UGV) simultaneous localization and Mapping (SLAM) system- NAVIS. Field tests are carried in an indoor parking place with NAVIS to evaluate the effectiveness of the proposed method. The result shows that all the small size objects like pedestrians can be detected and removed quickly; large size of objects like cars can be detected and removed partly.
\n
BibTeX
@inproceedings{yu2015approach,
  author = {Tong Yu and Jingbin Liu and Yuwei Chen and Jian Tang},
  title = {{An Approach of Dynamic Object Removing for Indoor Mapping Based on UGV SLAM}},
  booktitle = {SHILAP},
  year = {2015},
}

Loading related content...

Figure