Research map / Vision-language and multimodal / Robot action and navigation
FocusNav: Learning Task-directed Perception via Action-Aware Future Reconstruction in Vision-and-Language Navigation
Caifeng liu, Xueqiong Li, Chenghao Shi, weida chen, Yuhao Wang, Quan Zhang, Yuanxi Peng, Guijian Tang, Shaowu Yang
Read the paperCheck your idea against 137,217 papers
AI-extracted summary. Written by AI from the paper's abstract only. Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Proposes FocusNav, a vision-and-language navigation approach that learns task-directed perception via action-aware future reconstruction to keep agents focused on navigation-critical regions.
Problem
Vision-and-language navigation requires agents to maintain task-directed spatial perception on navigation-critical regions; existing hierarchical planners that select waypoints or subgoals often lose focus, producing inaccurate or redundant subgoals that mislead low-level control.
Approach
Learns task-directed perception through action-aware future reconstruction, training the agent to reconstruct future observations conditioned on its actions so that perception is focused on navigation-relevant content (details beyond the abstract are not stated in the provided text).
Closest papers by meaning
- LatentPilot: Scene-Aware Vision-and-Language Navigation by Dreaming Ahead with Latent Visual Reasoning (ECCV 2026 · similarity 0.91)
- AgentVLN: Towards Agentic Vision-and-Language Navigation (ECCV 2026 · similarity 0.91)
- Dive into the Scene: Breaking the Perceptual Bottleneck in Vision-Language Decision Making via Focus Plan Generation (ICML 2026 · similarity 0.90)
- Improving Vision-and-Language Navigation by Generating Future-View Image Semantics (CVPR 2023 · similarity 0.90)
- World–Value–Action Model: Implicit Planning for Vision–Language–Action Systems (NeurIPS 2026 · similarity 0.89)
- STRIDER: Navigation via Instruction-Aligned Structural Decision Space Optimization (NeurIPS 2025 · 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.
