人工智能论文英文版-Object Navigation with Structure-Semantic Reasoning-Based Multi-level Map and Multimodal Decision-Making LLM.pdf

ObjectNavigationwithStructure-Semantic

Reasoning-BasedMulti-levelMapandMultimodal

Decision-MakingLLM

ChongshangYanJiaxuanHe

SchoolofAutomationSchoolofAutomation

5BeijingInstituteofTechnologyBeijingInstituteofTechnology

23120240840@bit.edu.cn1120213032@bit.edu.cn

0

2DelunLiYiYang

nSchoolofAutomationSchoolofAutomation

uBeijingInstituteofTechnologyBeijingInstituteofTechnology

J3220231166@bit.edu.cnyang_yi@bit.edu.cn

6

]WenjieSong

OSchoolofAutomation

RBeijingInstituteofTechnology

.songwj@bit.edu.cn

s

c

[

1Abstract

v

6Thezero-shotobjectnavigation(ZSON)inunknownopen-endedenvironments

9coupledwithsemanticallynoveltargetoftensuffersfromthesignificantdecline

8inperformanceduetotheneglectofhigh-dimensionalimplicitsceneinformation

5andthelong-rangetargetsearchingtask.Toaddressthis,weproposedanactive

0

.objectnavigationframeworkwithEnvironmentalAttributesMap(EAM)and

6