Research map / Multi-agent systems and games / Human-AI and goal-conditioned
One Step at a Time: Long-Horizon Vision-and-Language Navigation With Milestones
Check 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 |
|---|---|---|---|
| Add M-Track to LSTM-L | LSTM-L · ALFRED Test Unseen | SR: 8.70 -> 13.28 | Helped |
| Add M-Track to VLN↻BERT-L | VLN↻BERT-L · ALFRED Test Unseen | SR: 12.23 -> 16.29 | Helped |
| Add reinforcement learning (ablation) | LSTM-L · ALFRED validation unseen | SR: 1.82 -> 8.03 | Helped |
| Add reinforcement learning (ablation) | VLN↻BERT-L · ALFRED validation unseen | SR: 3.66 -> 9.37 | Helped |
| Pre-train the object detector on ALFRED (ablation) | LSTM-L + RL · ALFRED validation unseen | SR: 8.03 -> 9.37 | Helped |
| Pre-train the object detector on ALFRED (ablation) | VLN↻BERT-L + RL · ALFRED validation unseen | SR: 9.37 -> 10.35 | Helped |
| Replace explicit milestone checking with a binary classifier (ablation) | LSTM-L + RL + ALFRED-OD · ALFRED validation unseen | SR: 15.22 -> 10.83 | Hurt |
| Replace explicit milestone checking with a binary classifier (ablation) | VLN↻BERT-L + RL + ALFRED-OD · ALFRED validation unseen | SR: 17.05 -> 14.85 | Hurt |
| Check milestones after rather than before actions (ablation) | LSTM-L + RL + ALFRED-OD · ALFRED validation unseen | SR: 15.83 -> 15.22 | Hurt |
| Check milestones after rather than before actions (ablation) | VLN↻BERT-L + RL + ALFRED-OD · ALFRED validation unseen | SR: 17.29 -> 17.05 | Hurt |
| Use ground-truth milestones instead of predicted milestones (ablation) | LSTM-L + M-Track · ALFRED validation unseen | SR: 15.83 -> 20.36 | Helped |
| Use ground-truth milestones instead of predicted milestones (ablation) | VLN↻BERT-L + M-Track · ALFRED validation unseen | SR: 17.29 -> 24.38 | Helped |
Closest papers by meaning
- STRIDER: Navigation via Instruction-Aligned Structural Decision Space Optimization (NeurIPS 2025 · similarity 0.86)
- A Subgoal-driven RL Framework for Improving Long-Horizon Web Agents (NeurIPS 2026 · similarity 0.86)
- Plan-and-Act: Improving Planning of Agents for Long-Horizon Tasks (ICML 2025 · similarity 0.86)
- NavGPT: Explicit Reasoning in Vision-and-Language Navigation with Large Language Models (AAAI 2024 · similarity 0.85)
- GoalLadder: Incremental Goal Discovery with Vision-Language Models (NeurIPS 2025 · similarity 0.85)
- HATS: Hardness-Aware Trajectory Synthesis for GUI Agents (CVPR 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.