AI Paper Atlas

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

Find What You Want: Learning Demand-conditioned Object Attribute Space for Demand-driven Navigation

Hongcheng Wang, Andy Guan Hong Chen, Xiaoqi Li, Mingdong Wu, Hao Dong

NeurIPS 2023 · Vision-language and multimodal › Robot action and navigation · approach: Reasoning-enhanced vision-language navigation · 2 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
remove BERT demand encoder (use MLP) (ablation)Ours · DDN Seen Scene/Unseen Ins.NSR: 19.3(3.3) -> 8.4(9.3)Hurt
remove BERT demand encoder (use MLP) (ablation)Ours · DDN Seen Scene/Seen Ins.NSR: 21.5(3.0) -> 12.2(12.2)Hurt
remove demand-conditioned contrastive pretraining (ablation)Ours · DDN Seen Scene/Seen Ins.NSR: 21.5(3.0) -> 13.1(3.5)Hurt
remove demand-conditioned contrastive pretraining (ablation)Ours · DDN Unseen Scene/Unseen Ins.NSR: 16.1(1.5) -> 11.8(0.58)Hurt
replace attribute transformer with MLP (ablation)Ours · DDN Seen Scene/Seen Ins.NSR: 21.5(3.0) -> 15.6(10.3)Hurt
replace attribute transformer with MLP (ablation)Ours · DDN Unseen Scene/Seen Ins.NSPL: 6.4(1.0) -> 4.5(2.3)Hurt
replace attribute transformer with MLP (ablation)Ours · DDN Seen Scene/Unseen Ins.SSR: 4.5(1.8) -> 7.2(2.3)No clear effect
use CLIP features instead of DETR featuresVTN-demand · DDN Seen Scene/Seen Ins.NSR: 6.3(1.7) -> 12.0(3.0)Improved
use GPT-3 parsing to map demand to objectsVTN · DDN Seen Scene/Seen Ins.NSR: 6.3(1.7) -> 1.6(1.1)Hurt
replace one-hot with BERT demand featuresVTN · DDN Seen Scene/Seen Ins.NSR: 5.2 -> 6.3(1.7)Improved
use CLIP alignment (ZSON-demand) vs randomZSON-demand · DDN Seen Scene/Seen Ins.NSR: 5.2(0.72) -> 4.2(1.2)Hurt
GPT-3 prompt navigation without visual input vs randomGPT-3-Prompt* · DDNNSR: 5.2(0.72) -> 0.3Hurt

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