Research map / Vision-language and multimodal / Robot action and navigation
Situational Fusion of Visual Representation for Visual Navigation
William B. Shen, Danfei Xu, Yuke Zhu, Leonidas J. Guibas, Li Fei-Fei, Silvio Savarese
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 changed | Where | Measured | Effect |
|---|---|---|---|
| use pretrained visual representations by concatenation instead of ResNet end-to-end | ResNet · Gibson navigation average | success rate (%): 20.9 -> 25.6 | Improved |
| fuse representations at action level instead of feature level (ablation) | feature-level fusion · Gibson navigation average | success rate (%): 23.1 -> 40.3 | Improved |
| add load-balancing loss (LBL) (ablation) | feature-level fusion · Gibson navigation average | success rate (%): 23.1 -> 25.5 | Improved |
| add load-balancing loss (LBL) (ablation) | action-level fusion · Gibson navigation average | success rate (%): 40.3 -> 39.8 | No clear effect |
| add inter-task affinity regularization (T.Aff) (ablation) | feature-level fusion · Gibson navigation average | success rate (%): 23.1 -> 26.0 | Improved |
| add inter-task affinity regularization (T.Aff) (ablation) | action-level fusion · Gibson navigation average | success rate (%): 40.3 -> 44.2 | Improved |
| add T.Aff on top of LBL (ablation) | feature-level fusion · Gibson navigation average | success rate (%): 25.5 -> 31.2 | Improved |
| add T.Aff on top of LBL (ablation) | action-level fusion · Gibson navigation average | success rate (%): 39.8 -> 44.0 | Improved |
| replace majority voting by learned action-level fusion with T.Aff | Maj. voting · Gibson navigation average | success rate (%): 41.1 -> 44.2 | Improved |
| add T.Aff to action-level fusion (ablation) | action-level fusion · Gibson navigation (go to table) | success rate (%): 28.9 -> 35.9 | Improved |
| add T.Aff to action-level fusion (ablation) | action-level fusion · Gibson navigation (go to chair) | success rate (%): 28.9 -> 36.3 | Improved |
| add T.Aff to action-level fusion (ablation) | action-level fusion · Gibson navigation (go to bed) | success rate (%): 48.8 -> 48.8 | No clear effect |
Closest papers by meaning
- Towards Versatile Embodied Navigation (NeurIPS 2022 · similarity 0.88)
- Towards Learning a Generic Agent for Vision-and-Language Navigation via Pre-Training (CVPR 2020 · similarity 0.87)
- Transferable Representation Learning in Vision-and-Language Navigation (ICCV 2019 · similarity 0.87)
- Seeing through Uncertainty: Robust Task-Oriented Optimization in Visual Navigation (NeurIPS 2025 · similarity 0.86)
- End-to-End (Instance)-Image Goal Navigation through Correspondence as an Emergent Phenomenon (ICLR 2024 · similarity 0.85)
- REGNav: Room Expert Guided Image-Goal Navigation (AAAI 2025 · similarity 0.85)
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.
