Research map / Vision-language and multimodal / Robot action and navigation
Interactive Grounded Language Acquisition and Generalization in a 2D World
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 |
|---|---|---|---|
| Share a concept detection function between language prediction and grounding | Proposed model · XWorld ZS2 | NAV success rate | Helped |
| Disentangle explicit language grounding from downstream computations | Proposed model · XWorld ZS1 | NAV success rate: 84.3% | Helped |
| Disentangle explicit language grounding from downstream computations | Proposed model · XWorld ZS2 | NAV success rate: 85.2% | Helped |
| Use contextual attention instead of the interpreter | Proposed model · XWorld ZS2 | NAV success rate: up to 80% | Hurt |
| Use stacked attention instead of the interpreter | Proposed model · XWorld ZS2 | NAV success rate: up to 80% | Hurt |
| Use contextual attention instead of the interpreter | Proposed model · XWorld ZS2 | QA accuracy: up to 35% | Hurt |
| Use stacked attention instead of the interpreter | Proposed model · XWorld ZS2 | QA accuracy: up to 35% | Hurt |
| Remove the QA pipeline and train only on NAV (ablation) | Proposed model · XWorld NAV training | accumulated discounted reward per session: -0.8 to -3.0 | Hurt |
| Remove spatial attention and attend to the whole image | VIS-LSTM · XWorld NAV | navigation success rate | Hurt |
| Remove spatial attention and attend to the whole image | ConcatEmbed · XWorld NAV | navigation success rate | Hurt |
Closest papers by meaning
- Learning to Model the World With Language (ICML 2024 · similarity 0.89)
- Distilling Internet-Scale Vision-Language Models into Embodied Agents (ICML 2023 · similarity 0.88)
- Just Ask: An Interactive Learning Framework for Vision and Language Navigation (AAAI 2020 · similarity 0.87)
- SILG: The Multi-domain Symbolic Interactive Language Grounding Benchmark (NeurIPS 2021 · similarity 0.87)
- Ground-Compose-Reinforce: Grounding Language in Agentic Behaviours using Limited Data (NeurIPS 2025 · similarity 0.87)
- Prismatic VLMs: Investigating the Design Space of Visually-Conditioned Language Models (ICML 2024 · similarity 0.86)
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.