Research map / Large language models / Language, safety and interpretability
Controllable Navigation Instruction Generation with Chain of Thought Prompting
Xianghao Kong, Jinyu Chen, Wenguan Wang, Hang Su, Xiaolin Hu, Yi Yang, Si Liu
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 style-mixed training (mix instruction styles) (ablation) | C-Instructor baseline · REVERIE val unseen | CIDEr: 0.347 -> 0.397 | Improved |
| add spatial topology modeling task (STMT) (ablation) | Baseline + SMT · REVERIE val unseen | CIDEr: 0.397 -> 0.490 | Improved |
| add Chain of Thought with Landmarks (CoTL) (ablation) | Baseline + SMT + STMT · REVERIE val unseen | CIDEr: 0.490 -> 0.545 | Improved |
| add CoTL (ablation) | Baseline + SMT + STMT · REVERIE val unseen | SPICE: 0.129 -> 0.141 | Improved |
| add CoTL (ablation) | Baseline + SMT + STMT · R2R val unseen | Rouge: 0.479 -> 0.473 | Mixed |
| add SMT (ablation) | C-Instructor · REVERIE val seen | CIDEr: 0.844 -> 0.805 | Hurt |
| add SMT (ablation) | C-Instructor · REVERIE val unseen | BLEU-4: 0.369 -> 0.419 | Improved |
| use multimodal trajectory encoder instead of captioning views with BLIP (ablation) | C-Instructor · REVERIE val unseen | CIDEr: 0.432 -> 0.545 | Improved |
| augment HAMT training data with C-Instructor instructions | HAMT · REVERIE val unseen | SR: 32.95 -> 34.25 | Improved |
| augment HAMT training data with speaker-model instructions (BT, EDrop, CCC) | HAMT · REVERIE val unseen | SR: 32.95 -> 29.65 | Hurt |
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.
- Lana — LANA: A Language-Capable Navigator for Instruction Following and Generation (CVPR 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
- Scene Map-based Prompt Tuning for Navigation Instruction Generation (CVPR 2025 · similarity 0.90)
- Run, Ruminate, and Regulate: A Dual-process Thinking System for Vision-and-Language Navigation (AAAI 2026 · similarity 0.86)
- STRIDER: Navigation via Instruction-Aligned Structural Decision Space Optimization (NeurIPS 2025 · similarity 0.86)
- Improving Instruction-Following in Language Models through Activation Steering (ICLR 2025 · similarity 0.85)
- NavGPT: Explicit Reasoning in Vision-and-Language Navigation with Large Language Models (AAAI 2024 · similarity 0.85)
- Think, Plan, Paint: Layout-Aware Reasoning for Controllable Image Generation in Unified Models (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.
