Research map / Vision-language and multimodal / Robot action and navigation
TrajRAG: Retrieving Geometric-Semantic Experience for Zero-Shot Object Navigation
Yiyao Wang, Sixian Zhang, Keming Zhang, Xinhang Song, Songjie Du, Shuqiang Jiang
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 |
|---|---|---|---|
| add topo-polar semantics to geometry (TPS-G + TPS-S) (ablation) | TrajRAG node representation · HM3Dv1 val | SR (%): 48.1 -> 61.7 | Improved |
| use topo-polar semantics instead of textual neighbor aggregation (ablation) | TrajRAG node representation · HM3Dv1 val | SR (%): 53.9 -> 57.3 | Improved |
| use TPS-G and TPS-S instead of TNA (ablation) | TrajRAG · HM3Dv1 val | SPL (%): 25.7 -> 33.2 | Improved |
| add coarse scene-level retrieval before fine retrieval (ablation) | TrajRAG · HM3Dv1 val | SR (%): 54.3 -> 61.7 | Improved |
| use trained sequence encoder instead of pre-trained text embedding (ablation) | TrajRAG fine retrieval · HM3Dv1 val | SR (%): 57.8 -> 61.7 | Improved |
| use trajectory retrieval instead of TrajTextRAG | TrajRAG · HM3Dv1 val | SR (%): 53.3 -> 61.7 | Improved |
| use trajectory retrieval instead of GraphRAG | TrajRAG · HM3Dv1 val | SR (%): 55.2 -> 61.7 | Improved |
| combine HM3Dv1 and MP3D training corpora (ablation) | TrajRAG · MP3D val | SR (%): 39.4 -> 42.6 | Improved |
| combine HM3Dv1 and MP3D training corpora (ablation) | TrajRAG · HM3Dv1 val | SR (%): 61.7 -> 62.5 | Improved |
| use in-domain MP3D corpus over HM3Dv1 corpus (ablation) | TrajRAG · MP3D val | SR (%): 39.4 -> 41.7 | Improved |
Closest papers by meaning
- MemoNav: Working Memory Model for Visual Navigation (CVPR 2024 · similarity 0.88)
- Scene Map-based Prompt Tuning for Navigation Instruction Generation (CVPR 2025 · similarity 0.87)
- BeliefMapNav: 3D Voxel-Based Belief Map for Zero-Shot Object Navigation (NeurIPS 2025 · similarity 0.87)
- LangMap: A Human-Verified Benchmark for Hierarchical Open-Vocabulary Goal Navigation (NeurIPS 2026 · similarity 0.87)
- NavBench: Probing Multimodal Large Language Models for Embodied Navigation (NeurIPS 2025 · similarity 0.87)
- STRIDER: Navigation via Instruction-Aligned Structural Decision Space Optimization (NeurIPS 2025 · similarity 0.86)
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.
