Research map / Vision-language and multimodal / Robot action and navigation
Scene-Intuitive Agent for Remote Embodied Visual Grounding
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 |
|---|---|---|---|
| Use the Scene Grounding-pretrained ViL encoder instead of a raw ViL encoder (ablation) | REVERIE navigation agent · REVERIE Val Seen and Val Unseen | RG SPL: 21.94 -> 33.92; 4.84 -> 7.31 | Helped |
| Replace MattNet with ViLPointer (ablation) | REVERIE navigation agent · REVERIE Val Seen and Val Unseen | Navigation Length: 14.56 -> 13.71; 42.62 -> 37.31 | Helped |
| Replace the conventional policy with the memory-augmented policy (ablation) | REVERIE navigation agent · REVERIE Val Seen and Val Unseen | RGS: 23.82 -> 24.88; 8.89 -> 11.08 | Helped |
| Use memory and state block counts (3, 3) rather than (1, 1) (ablation) | Memory-augmented action decoder · REVERIE Val Seen and Val Unseen | RG SPL: 38.76 -> 42.65; 9.19 -> 11.56 | Helped |
| Increase memory and state block counts from (3, 3) to (5, 5) (ablation) | Memory-augmented action decoder · REVERIE Val Seen | RG SPL: 42.65 -> 41.10 | Hurt |
| Fuse action, scene-grounding, and object-grounding logits (ablation) | REVERIE navigation agent · REVERIE Val Seen and Val Unseen | RG SPL: 42.19 -> 42.65; 10.54 -> 11.56 | Helped |
| Use viewpoint-based rather than image-based object grounding (ablation) | ViLPointer · REVERIE Val Seen and Val Unseen | object grounding accuracy: 65.72 -> 73.26; 55.53 -> 67.45 | Helped |
| Use the complete scene-intuitive agent rather than REVERIE baseline | REVERIE · REVERIE Val Seen, Val Unseen, and Test (Unseen) | RG SPL: 29.66 -> 42.65; 4.67 -> 11.56; 6.08 -> 9.20 | Helped |
Closest papers by meaning
- ESCA: Contextualizing Embodied Agents via Scene-Graph Generation (NeurIPS 2025 · similarity 0.88)
- EmbodiedBench: Comprehensive Benchmarking Multi-modal Large Language Models for Vision-Driven Embodied Agents (ICML 2025 · similarity 0.87)
- Augmented Commonsense Knowledge for Remote Object Grounding (AAAI 2024 · similarity 0.87)
- SceneWeaver: All-in-One 3D Scene Synthesis with an Extensible and Self-Reflective Agent (NeurIPS 2025 · similarity 0.86)
- Connecting the Dots: Training-Free Visual Grounding via Agentic Reasoning (AAAI 2026 · similarity 0.86)
- DenseGrounding: Improving Dense Language-Vision Semantics for Ego-centric 3D Visual Grounding (ICLR 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.