Research map / Vision-language and multimodal / Robot action and navigation
Talk2Nav: Long-Range Vision-and-Language Navigation with Dual Attention and Spatial Memory
Arun Balajee Vasudevan, Dengxin Dai, Luc Van Gool
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 explicit top-view memory image instead of no memory (ablation) | Talk2Nav agent · Talk2Nav (test) | SPL (%): 6.87 -> 11.92 | Improved |
| add top-view memory image instead of GPS coordinate sequence (ablation) | Talk2Nav agent · Talk2Nav (test) | SPL (%): 9.04 -> 11.92 | Improved |
| add GPS coordinate memory instead of no memory (ablation) | Talk2Nav agent · Talk2Nav (test) | SPL (%): 6.87 -> 9.04 | Improved |
| use ResNet features instead of pretrained SphereNet (ablation) | Talk2Nav agent · Talk2Nav (test) | SPL (%): 9.37 -> 11.92 | Improved |
| add Indicator Controller (ACT) instead of no IC (ablation) | Talk2Nav agent · Talk2Nav (test) | SPL (%): 9.71 -> 11.92 | Improved |
| use ACT instead of regular LSTM in the IC (ablation) | Talk2Nav agent · Talk2Nav (test) | SPL (%): 8.95 -> 11.92 | Improved |
| use generative matching instead of discriminative matching (ablation) | Talk2Nav agent · Talk2Nav (test) | SPL (%): 11.15 -> 11.92 | Improved |
| train with routes of all difficulty levels instead of 4-landmark routes only (ablation) | Talk2Nav agent · Talk2Nav (All landmark counts) | SPL (%): 49.93 -> 51.37 | Improved |
| train on 3-landmark routes instead of 4-landmark routes, test on 4 landmarks (ablation) | Talk2Nav agent · Talk2Nav (4 landmark routes) | SPL (%): 11.92 -> 11.51 | Hurt |
| train on 1-landmark routes instead of 4-landmark, test on 4 landmarks (ablation) | Talk2Nav agent · Talk2Nav (4 landmark routes) | SPL (%): 11.92 -> 8.45 | Hurt |
| Ours method vs best baseline Self-Monitoring | Talk2Nav agent · Talk2Nav (test) | SPL (%): 9.56 -> 11.92 | Improved |
| Ours method vs best baseline Self-Monitoring | Talk2Nav agent · Talk2Nav (test) | Navigation Error (m): 740.12 -> 633.85 | Improved |
Closest papers by meaning
- NavGPT: Explicit Reasoning in Vision-and-Language Navigation with Large Language Models (AAAI 2024 · similarity 0.89)
- Scene Map-based Prompt Tuning for Navigation Instruction Generation (CVPR 2025 · similarity 0.89)
- NavBench: Probing Multimodal Large Language Models for Embodied Navigation (NeurIPS 2025 · similarity 0.89)
- Run, Ruminate, and Regulate: A Dual-process Thinking System for Vision-and-Language Navigation (AAAI 2026 · similarity 0.88)
- CitySeeker: How Do VLMs Explore Embodied Urban Navigation with Implicit Human Needs? (ICLR 2026 · similarity 0.88)
- VELMA: Verbalization Embodiment of LLM Agents for Vision and Language Navigation in Street View (AAAI 2024 · similarity 0.88)
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.
