AI Paper Atlas

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

JanusVLN: Decoupling Semantics and Spatiality with Dual Implicit Memory for Vision-Language Navigation

Shuang Zeng, Dekang Qi, Xinyuan Chang, Feng Xiong, Xie Shichao, Xiaolong Wu, Shiyi Liang, Mu Xu, Xing Wei

ICLR 2026 · Vision-language and multimodal › Robot action and navigation · approach: Reasoning-enhanced vision-language navigation

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
add spatial implicit memory (VGGT KV cache) (ablation)JanusVLN · R2R-CE Val-UnseenSPL: 40.9 -> 49.2Improved
add semantic implicit memory (ablation)JanusVLN · R2R-CE Val-UnseenSR: 45.5 -> 52.8Improved
add both implicit memories (ablation)JanusVLN · R2R-CE Val-UnseenSR: 24.8 -> 52.8Improved
add extra DINOv2 or SigLIP 2 encoder (ablation)JanusVLN · R2R-CE Val-UnseenSR: 47.0 -> 47.5 / 47.9No clear effect
add randomly initialized VGGT (ablation)JanusVLN · R2R-CE Val-UnseenSR: 47.0 -> 47.2No clear effect
add pretrained VGGT as spatial encoder (ablation)JanusVLN · R2R-CE Val-UnseenSR: 47.0 -> 52.8Improved
cache KV history instead of re-running VGGT on all frames (ablation)JanusVLN · R2R-CE Val-Unseeninference time: 1549 ms -> 149 ms at 32 framesImproved
keep initial-window KV as attention sinks (ablation)JanusVLN · R2R-CE Val-UnseenSR: 51.0 -> 52.8Improved
increase memory size from 48 to 64 frames (ablation)JanusVLN · R2R-CE Val-UnseenSR: 52.8 -> 52.3No clear effect
spatial feature weight lambda 0.2 versus 0.5 or 0.1 (ablation)JanusVLN · R2R-CE Val-UnseenSR: 50.4 / 50.2 -> 52.8Improved
weighted addition fusion instead of concat (ablation)JanusVLN · R2R-CE Val-UnseenSR: 49.4 -> 52.8Improved
add ScaleVLN and DAgger data (ablation)JanusVLN · R2R-CE Val-UnseenSR: 52.8 -> 60.5Improved

Later papers that compared with this method

1 later accepted paper reports JanusVLN as a baseline in their results tables; 1 of them report a better number on most of the results they share with it.

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