Research map / Vision-language and multimodal / Robot action and navigation
Cross from Left to Right Brain: Adaptive Text Dreamer for Vision-and-Language Navigation
Pingrui Zhang, Yifei Su, Pengyuan Wu, Dong An, Li Zhang, Zhigang Wang, Dong Wang, Bin Zhao
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 both State Estimation LLM and Imagination LLM to the baseline (ablation) | DUET-style baseline · R2R val unseen | SR: 63 -> 75 | Improved |
| add both State Estimation LLM and Imagination LLM to the baseline (ablation) | DUET-style baseline · R2R val unseen | SPL: 52 -> 63 | Improved |
| add only one of the two LLM modules (SELLM or IMLLM) to the baseline (ablation) | DUET-style baseline · R2R val unseen | SR: 63 -> 72 / 73 | Improved |
| use both modules together instead of either one alone (ablation) | ATD · R2R val seen | SR: 71 -> 76 | Improved |
| ground imagination with state estimation (state as query) rather than reverse roles (ablation) | ATD SGCA · R2R val seen | SR: 71.79 -> 75.61 | Improved |
| ground imagination with state estimation (state as query) rather than reverse roles (ablation) | ATD SGCA · R2R val unseen | SR: 73.95 -> 74.63 | Improved |
| add SGCA cross-attention instead of parallel summation (ablation) | ATD · R2R val unseen | SPL: 61.89 -> 63.35 | Improved |
| increase SGCA layers from 1 to 4 (ablation) | ATD · R2R val seen | SPL: 63.15 -> 67.49 | Improved |
| increase SGCA layers from 1 to 4 (ablation) | ATD · R2R val unseen | SR: 74.54 -> 74.63 | No clear effect |
| use ATD instead of single-brain NavGPT2 (1.5B) | LLM-based VLN · R2R test unseen | SR: 71 -> 74 | Improved |
| keep LLM text generation ability (frozen 1.5B) instead of fully fine-tuned 7B NaviLLM | LLM-based VLN · R2R test unseen | SR: 68 -> 74 | Improved |
| use ATD instead of DUET in zero-shot transfer from R2R | navigation policy · REVERIE val unseen | SR: 24 -> 27 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- DUET — DUET: Cross-Modal Semantic Grounding for Contrastive Zero-Shot Learning (AAAI 2023) · better on 4 of 4 shared results
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- Planning from Imagination: Episodic Simulation and Episodic Memory for Vision-and-Language Navigation (AAAI 2025 · similarity 0.89)
- Using Left and Right Brains Together: Towards Vision and Language Planning (ICML 2024 · similarity 0.88)
- Run, Ruminate, and Regulate: A Dual-process Thinking System for Vision-and-Language Navigation (AAAI 2026 · similarity 0.88)
- Do Visual Imaginations Improve Vision-and-Language Navigation Agents? (CVPR 2025 · similarity 0.88)
- FantasyVLN: Unified Multimodal Chain-of-Thought Reasoning for Vision-and-Language Navigation (CVPR 2026 · similarity 0.88)
- ImagineNav: Prompting Vision-Language Models as Embodied Navigator through Scene Imagination (ICLR 2025 · similarity 0.87)
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.
