Research map / Vision-language and multimodal / Robot action and navigation
Zero Experience Required: Plug & Play Modular Transfer Learning for Semantic Visual Navigation
Check your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 |
|---|---|---|---|
| Add the view reward and view augmentation to image-goal policy training (ablation) | Image-goal navigation policy · Gibson split A | Succ.: 22.0 -> 29.2; 24.4 -> 29.2 | Helped |
| Add the view reward and view augmentation to image-goal policy training (ablation) | Image-goal navigation policy · Gibson split A | SPL: 18.8 -> 21.6; 17.3 -> 21.6 | Helped |
| Use the modular semantic-search transfer model rather than the strongest listed baseline | Image-goal navigation · Gibson split A | Succ.: 22.6 -> 29.2 | Helped |
| Use the modular semantic-search transfer model rather than the strongest listed baseline | Image-goal navigation · Gibson split B | SPL: 12.4 -> 23.6 | Helped |
| Transfer the full model without target-task interactions | Semantic navigation tasks · ObjectNav-Label | success rate: 8.0 -> 11.3 | Helped |
| Transfer the full model and fine-tune on the target task | Semantic navigation tasks · ObjectNav-Label | success rate: 13.9 -> 21.9 | Helped |
| Transfer the full model rather than only the observation encoder (ablation) | ObjectNav-Label · ObjectNav-Label | success rate | Helped |
| Increase source-task experience (ablation) | Semantic search policy · Downstream ZSEL tasks | success rate | Helped |
| Increase vision resolution from 128 to 256 (ablation) | ImageNav source policy evaluated on ObjectNav-Label · ObjectNav-Label ZSEL | success rate | Helped |
Closest papers by meaning
- Unsupervised Reinforcement Learning of Transferable Meta-Skills for Embodied Navigation (CVPR 2020 · similarity 0.88)
- Learning to Learn How to Learn: Self-Adaptive Visual Navigation Using Meta-Learning (CVPR 2019 · similarity 0.87)
- RAO-Nav: Probing Omni-Language Models for Zero-shot Semantic Audio-Visual Navigation (NeurIPS 2026 · similarity 0.86)
- Prompt-based Visual Alignment for Zero-shot Policy Transfer (ICML 2024 · similarity 0.86)
- End-to-End (Instance)-Image Goal Navigation through Correspondence as an Emergent Phenomenon (ICLR 2024 · similarity 0.85)
- RILA: Reflective and Imaginative Language Agent for Zero-Shot Semantic Audio-Visual Navigation (CVPR 2024 · 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.