Semantic maps from multiple visual cues

作者：

Highlights：

• Semantic mapping based on place and object recognition strategies.

• Place recognition involves learning through bag of visual words.

• Object recognition relies on HTM deep learning.

• Decision making endorses time embedded voting for objects and places.

• Topometric map is associated with place belief distribution.

摘要

•Semantic mapping based on place and object recognition strategies.•Place recognition involves learning through bag of visual words.•Object recognition relies on HTM deep learning.•Decision making endorses time embedded voting for objects and places.•Topometric map is associated with place belief distribution.

论文关键词：Semantic map,3D object map,3D topological map,Place classification,Object recognition,SLAM

论文评审过程：Received 14 March 2016, Revised 9 October 2016, Accepted 10 October 2016, Available online 11 October 2016, Version of Record 18 October 2016.

论文官网地址：https://doi.org/10.1016/j.eswa.2016.10.014