Research map / Vision-language and multimodal / Robot action and navigation
NaVLA$^2$: A Vision-Language-Audio-Action Model for Multimodal Instruction Navigation
Jugang Fan, Peihao Chen, Changhao Li, Qing Du, Jian Chen, Mingkui Tan
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 spatial audio branch to audio encoder (no audio branches baseline) (ablation) | NaVLA2 · MINav-bench (test-mini) | SR: 0.206 -> 0.258 | Improved |
| add semantic audio branch to audio encoder (no audio branches baseline) (ablation) | NaVLA2 · MINav-bench (test-mini) | SR: 0.206 -> 0.219 | Improved |
| add semantic branch on top of spatial branch (ablation) | NaVLA2 · MINav-bench (test-mini) | SR: 0.258 -> 0.272 | Improved |
| add semantic branch on top of spatial branch (ablation) | NaVLA2 · MINav-bench (test-mini) | SPL: 0.201 -> 0.196 | No clear effect |
| add spatial branch on top of semantic branch (ablation) | NaVLA2 · MINav-bench (test-mini) | SPL: 0.168 -> 0.196 | Improved |
| decode actions from [NAV] embedding with action head instead of LLM text actions (ablation) | NaVLA2 · MINav-bench (test-mini) | SPL: 0.165 -> 0.196 | Improved |
| decode actions from [NAV] embedding with action head instead of LLM text actions (ablation) | NaVLA2 · MINav-bench (test-mini) | SR: 0.256 -> 0.272 | Improved |
| add chain-of-thought before [NAV] token (ablation) | NaVLA2 · MINav-bench (test-mini) | SR: 0.261 -> 0.272 | Improved |
| add chain-of-thought before [NAV] token (ablation) | NaVLA2 · MINav-bench (test-mini) | SPL: 0.192 -> 0.196 | Improved |
| give full multimodal instruction with landmarks instead of goal object only | RL baseline · MINav-bench (test-mini) | SR: 0.064 -> 0.156 | Improved |
| use VLA with spatial audio and CoThinkAct instead of Navid | NaVLA2 vs Navid · MINav-bench (test-mini) | SR: 0.131 -> 0.272 | Improved |
| use more modality types in an episode's references instead of a single modality | NaVLA2 · MINav-bench (test-mini) | SR | Improved |
Closest papers by meaning
- NavBench: Probing Multimodal Large Language Models for Embodied Navigation (NeurIPS 2025 · similarity 0.92)
- RAO-Nav: Probing Omni-Language Models for Zero-shot Semantic Audio-Visual Navigation (NeurIPS 2026 · similarity 0.91)
- NavGPT: Explicit Reasoning in Vision-and-Language Navigation with Large Language Models (AAAI 2024 · similarity 0.91)
- RILA: Reflective and Imaginative Language Agent for Zero-Shot Semantic Audio-Visual Navigation (CVPR 2024 · similarity 0.91)
- CAVEN: An Embodied Conversational Agent for Efficient Audio-Visual Navigation in Noisy Environments (AAAI 2024 · similarity 0.90)
- EmbodiedBench: Comprehensive Benchmarking Multi-modal Large Language Models for Vision-Driven Embodied Agents (ICML 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.
