AI Paper Atlas

Research map / Vision-language and multimodal / Robot action and navigation

ETPNav: Evolving Topological Planning for Vision-Language Navigation in Continuous Environments

Dong Un An, Hanqing Wang, Wenguan Wang, Zun Wang, Yan Huang, Keji He, Liang Wang

TPAMI 2024 · Vision-language and multimodal › Robot action and navigation · approach: Pretraining and data augmentation for VLN · 154 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
use depth-only instead of RGB-only for waypoint prediction (ablation)ETPNav · R2R-CE Val UnseenSR: 51.66 -> 57.21Improved
use depth-only instead of RGB+depth for waypoint prediction (ablation)ETPNav · R2R-CE Val UnseenSR / SPL: 56.44 -> 57.21 SR, 48.53 -> 49.15 SPLImproved
add waypoint accumulation (Acc.) for ghost nodes (ablation)ETPNav · R2R-CE Val UnseenSR / SPL: 55.89 -> 57.21 SR, 47.92 -> 49.15 SPLImproved
add ghost node deletion (Del.) (ablation)ETPNav · R2R-CE Val UnseenSR / SPL: 52.41 -> 57.21 SR, 45.02 -> 49.15 SPLImproved
raise localization threshold gamma from 0.25 m to 1.00 m (ablation)ETPNav · R2R-CE Val UnseenSR: 56.71 -> 42.30Hurt
use global planning space instead of local (ablation)ETPNav · R2R-CE Val UnseenSR: 53.92 -> 57.21Improved
add graph-aware self-attention (GASA) (ablation)ETPNav global planning · R2R-CE Val UnseenSR: 55.97 -> 57.21Improved
add MLM pre-training over no initialization (ablation)ETPNav · R2R-CE Val UnseenSR / SPL: 37.41 -> 48.23 SR, 30.28 -> 37.73 SPLImproved
add SAP pre-training over no initialization (ablation)ETPNav · R2R-CE Val UnseenSR / SPL: 37.41 -> 52.37 SR, 30.28 -> 42.77 SPLImproved
combine MLM and SAP pre-training (ablation)ETPNav · R2R-CE Val UnseenSR: 52.37 -> 57.21Improved
pre-train on Habitat RGB instead of MP3D RGB (ablation)ETPNav · R2R-CE Val UnseenSR / SPL: 53.24 -> 55.92 SR, 44.09 -> 46.97 SPLImproved
add depth input to pre-training (ablation)ETPNav · R2R-CE Val UnseenSR / SPL: 55.92 -> 57.21 SR, 46.97 -> 49.15 SPLImproved

Later papers that compared with this method

15 later accepted papers report ETPNav as a baseline in their results tables; 11 of them report a better number on most of the results they share with it.

7 more not listed.

Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy