Research map / Multi-agent systems and games / Human-AI and goal-conditioned
Visual Graph Memory With Unsupervised Representation for Visual Navigation
Obin Kwon, Nuri Kim, Yunho Choi, Hwiyeon Yoo, Jeongho Park, Songhwai Oh
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 |
|---|---|---|---|
| replace stacked memory (SMT) with graph memory VGM | SMT · Gibson image-goal navigation (Overall) | success rate: 0.68 -> 0.76 | Improved |
| replace stacked memory (SMT) with graph memory VGM | SMT · Gibson image-goal navigation (Overall) | SPL: 0.56 -> 0.64 | Improved |
| replace metric map (Exp4nav) with VGM | Exp4nav · Gibson image-goal navigation (Overall) | success rate: 0.59 -> 0.76 | Improved |
| replace metric map (Exp4nav) with VGM | Exp4nav · Gibson image-goal navigation (Overall) | SPL: 0.51 -> 0.64 | Improved |
| use VGM instead of SMT stored observations | SMT · Gibson image-goal navigation | memory elements: 96% fewer memory elements | Improved |
| lower similarity threshold s_th from 0.75 to 0.30 (ablation) | VGM · Gibson hard episodes | success rate: 0.61 -> 0.60 | No clear effect |
| lower similarity threshold s_th from 0.75 to 0.30 (ablation) | VGM · Gibson hard episodes | SPL: 0.46 -> 0.41 | Hurt |
| lower similarity threshold s_th from 0.75 to 0.30 (ablation) | VGM · Gibson hard episodes | localization accuracy: 0.94 -> 0.97 | No clear effect |
| add pose sensor noise to metric-map and pose-based baselines | ANS and Neural Planner · Gibson hard episodes | success rate | Hurt |
| add pose sensor noise to VGM | VGM · Gibson hard episodes | success rate | No clear effect |
| add auxiliary losses (ablation) | VGM · Gibson | success rate | Improved |
| train F_loc with supervised learning instead of unsupervised (ablation) | VGM · Gibson | success rate: about 8% improvement | Improved |
Closest papers by meaning
- MemoNav: Working Memory Model for Visual Navigation (CVPR 2024 · similarity 0.90)
- VTNet: Visual Transformer Network for Object Goal Navigation (ICLR 2021 · similarity 0.88)
- Frontier-enhanced Topological Memory with Improved Exploration Awareness for Embodied Visual Navigation (ECCV 2024 · similarity 0.87)
- Pose Invariant Topological Memory for Visual Navigation (ICCV 2021 · similarity 0.87)
- Hierarchical Semantic-Augmented Navigation: Optimal Transport and Graph-Driven Reasoning for Vision-Language Navigation (NeurIPS 2025 · similarity 0.86)
- Change-Robust Online Topological Memory for Long-Term Relocalization and Semantic Navigation (NeurIPS 2026 · 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.
