Research map / Vision-language and multimodal / Robot action and navigation
CapNav: Benchmarking Vision Language Models on Capability-conditioned Indoor Navigation
Xia Su, Ruiqi Chen, Benlin Liu, Jingwei Ma, Zonglin Di, Ranjay Krishna, Jon Froehlich
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 |
|---|---|---|---|
| enable thinking mode (9 matched pairs) (ablation) | VLMs on CapNav · CapNav | CapNav score: +6.87% | Improved |
| enable thinking mode (ablation) | VLMs on CapNav · CapNav | inference time: 14.94 s -> 123.94 s | Hurt |
| enable thinking mode (ablation) | Intern-VL3.5-8B · CapNav | CapNav score: 41.72 -> 53.72 | Improved |
| enable thinking mode (ablation) | Keye-VL-1.5-8B · CapNav | CapNav score: 47.28 -> 50.47 | Improved |
| enable thinking mode (ablation) | Doubao-Seed-1.6 · CapNav | CapNav score: 61.91 -> 62.12 | No clear effect |
| enable thinking mode (ablation) | Intern-VL3.5-14B · CapNav | reasoning validity: 25.18 -> 24.97 | No clear effect |
| increase input frames (16, 32, 64, 1 FPS) (ablation) | VLMs on CapNav · CapNav | CapNav score | Mixed |
| LoRA fine-tune on augmented 13K tasks | Qwen3-VL-8B-Instruct · CapNav | CapNav score: 45.26% -> 55.18% | Improved |
| LoRA fine-tune on augmented 13K tasks | Qwen3-VL-8B-Instruct · CapNav | reasoning validity: 0.30 -> 0.25 | Hurt |
| tighten mobility constraints (humanoid vs adult human) | VLMs on CapNav · CapNav | mean CapNav score: 57.83% -> 39.12% | Hurt |
| use spatial-reasoning models instead of general VLMs | Spatial-MLLM-4B · CapNav | CapNav score: 30.15 | Hurt |
| vary CapNav score weights (0.5 on one component) (ablation) | model ranking · CapNav | Kendall tau: 0.909 | No clear effect |
Closest papers by meaning
- Human-Aware Vision-and-Language Navigation: Bridging Simulation to Reality with Dynamic Human Interactions (NeurIPS 2024 · similarity 0.90)
- Hierarchical Semantic-Augmented Navigation: Optimal Transport and Graph-Driven Reasoning for Vision-Language Navigation (NeurIPS 2025 · similarity 0.90)
- NavBench: Probing Multimodal Large Language Models for Embodied Navigation (NeurIPS 2025 · similarity 0.89)
- Seeing Across Views: Benchmarking Spatial Reasoning of Vision-Language Models in Robotic Scenes (ICLR 2026 · similarity 0.89)
- ImagineNav: Prompting Vision-Language Models as Embodied Navigator through Scene Imagination (ICLR 2025 · similarity 0.89)
- PhysVLM: Enabling Visual Language Models to Understand Robotic Physical Reachability (CVPR 2025 · similarity 0.89)
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.
