Research map / Multi-agent systems and games / Human-AI and goal-conditioned
Just Ask: An Interactive Learning Framework for Vision and Language Navigation
Read the paperCheck 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 |
|---|---|---|---|
| Enable interaction through confusion-triggered questions | Base Model · R2R val_unseen | success rate: 0.471 -> 0.807 | Helped |
| Increase MC confusion threshold from ε=0.1 to ε=0.5 (ablation) | MC · R2R val_unseen | success rate: 0.562 -> 0.807 | Helped |
| Increase the ASA ask penalty from r_ask=0.1 to r_ask=0.5 (ablation) | ASA · R2R val_unseen | success rate: 0.732 -> 0.494 | Hurt |
| Add deviation shaping to the ASA reward (ablation) | ASA · R2R navigation | number of questions | Helped |
| Increase simulated response noise from C=0.0 to C=0.4 (ablation) | ASA-0.1 · R2R val_unseen | success rate: 0.732 -> 0.722 | No clear effect |
| Increase simulated response noise from C=0.0 to C=0.4 (ablation) | MC-0.5 · R2R val_unseen | success rate: 0.807 -> 0.562 | Hurt |
| Fine-tune using human-guided interaction data instead of pre-exploration data | Navigation agent · R2R val_unseen, disjoint split | success rate: 0.504 -> 0.554 | Helped |
| Use supervised-only fine-tuning instead of mixed supervised and RL training (ablation) | Human-Guided Exploration · R2R val_unseen | success rate: 0.514 vs. 0.483 | Mixed |
| Use human-guided data augmentation rather than pre-exploration | Navigation agent · R2R val_unseen | success rate: 0.519 vs. 0.472 | Helped |
Closest papers by meaning
- GoalLadder: Incremental Goal Discovery with Vision-Language Models (NeurIPS 2025 · similarity 0.89)
- Test-Time Adaptation for Online Vision-Language Navigation with Feedback-based Reinforcement Learning (ICML 2025 · similarity 0.88)
- Aux-Think: Exploring Reasoning Strategies for Data-Efficient Vision-Language Navigation (NeurIPS 2025 · similarity 0.88)
- Active Test-time Vision-Language Navigation (NeurIPS 2025 · similarity 0.88)
- NavGPT: Explicit Reasoning in Vision-and-Language Navigation with Large Language Models (AAAI 2024 · similarity 0.87)
- SenseSearch: Empowering Vision-Language Models with High-Resolution Agentic Search-Reasoning via Reinforcement Learning (CVPR 2026 · 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.