Context-Aware Replanning with Pre-Explored Semantic Map for Object Navigation

Po-Chen Ko, Hung-Ting Su, CY Chen, Jia-Fong Yeh, Min Sun, Winston H. Hsu
Proceedings of The 8th Conference on Robot Learning, PMLR 270:4253-4267, 2025.

Abstract

Pre-explored Semantic Map, constructed through prior exploration using visual language models (VLMs), has proven effective as a foundational element for training-free robotic applications. However, existing approaches assume the map’s accuracy and do not provide effective mechanisms for revising decisions based on incorrect maps. This work introduces Context-Aware Replanning (CARe),, which estimates map uncertainty through confidence scores and multi-view consistency, enabling the agent to revise erroneous decisions stemming from inaccurate maps without additional labels. We demonstrate the effectiveness of our proposed method using two modern map backbones, VLMaps and OpenMask3D, and show significant improvements in performance on object navigation tasks.

Cite this Paper

BibTeX
@InProceedings{pmlr-v270-ko25b, title = {Context-Aware Replanning with Pre-Explored Semantic Map for Object Navigation}, author = {Ko, Po-Chen and Su, Hung-Ting and Chen, CY and Yeh, Jia-Fong and Sun, Min and Hsu, Winston H.}, booktitle = {Proceedings of The 8th Conference on Robot Learning}, pages = {4253--4267}, year = {2025}, editor = {Agrawal, Pulkit and Kroemer, Oliver and Burgard, Wolfram}, volume = {270}, series = {Proceedings of Machine Learning Research}, month = {06--09 Nov}, publisher = {PMLR}, pdf = {https://raw.githubusercontent.com/mlresearch/v270/main/assets/ko25b/ko25b.pdf}, url = {https://proceedings.mlr.press/v270/ko25b.html}, abstract = {Pre-explored Semantic Map, constructed through prior exploration using visual language models (VLMs), has proven effective as a foundational element for training-free robotic applications. However, existing approaches assume the map’s accuracy and do not provide effective mechanisms for revising decisions based on incorrect maps. This work introduces Context-Aware Replanning (CARe),, which estimates map uncertainty through confidence scores and multi-view consistency, enabling the agent to revise erroneous decisions stemming from inaccurate maps without additional labels. We demonstrate the effectiveness of our proposed method using two modern map backbones, VLMaps and OpenMask3D, and show significant improvements in performance on object navigation tasks.} }
Endnote
%0 Conference Paper %T Context-Aware Replanning with Pre-Explored Semantic Map for Object Navigation %A Po-Chen Ko %A Hung-Ting Su %A CY Chen %A Jia-Fong Yeh %A Min Sun %A Winston H. Hsu %B Proceedings of The 8th Conference on Robot Learning %C Proceedings of Machine Learning Research %D 2025 %E Pulkit Agrawal %E Oliver Kroemer %E Wolfram Burgard %F pmlr-v270-ko25b %I PMLR %P 4253--4267 %U https://proceedings.mlr.press/v270/ko25b.html %V 270 %X Pre-explored Semantic Map, constructed through prior exploration using visual language models (VLMs), has proven effective as a foundational element for training-free robotic applications. However, existing approaches assume the map’s accuracy and do not provide effective mechanisms for revising decisions based on incorrect maps. This work introduces Context-Aware Replanning (CARe),, which estimates map uncertainty through confidence scores and multi-view consistency, enabling the agent to revise erroneous decisions stemming from inaccurate maps without additional labels. We demonstrate the effectiveness of our proposed method using two modern map backbones, VLMaps and OpenMask3D, and show significant improvements in performance on object navigation tasks.
APA
Ko, P., Su, H., Chen, C., Yeh, J., Sun, M. & Hsu, W.H.. (2025). Context-Aware Replanning with Pre-Explored Semantic Map for Object Navigation. Proceedings of The 8th Conference on Robot Learning, in Proceedings of Machine Learning Research 270:4253-4267 Available from https://proceedings.mlr.press/v270/ko25b.html.

Related Material