Research map / Vision-language and multimodal / Robot action and navigation
FlowDec: Temporal Conditional Flow Decorruptor for Robust Continuous Vision-Language Navigation
Yufei Zhang, Changhao Chen
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 |
|---|---|---|---|
| apply FlowDec decorruption to NaVid | NaVid · R2R-CE (avg of 6 corruptions) | SR (%): 21.58 -> 27.04 | Improved |
| apply FlowDec decorruption to NaVid | NaVid · RxR-CE (avg of 6 corruptions) | SR (%): 30.75 -> 33.63 | Improved |
| apply FlowDec decorruption to NaVid | NaVid · R2R-CE (avg of 6 corruptions) | SPL (%): 19.11 -> 23.62 | Improved |
| apply FlowDec instead of Dec-CM | NaVid · R2R-CE (avg of 6 corruptions) | SR (%): 24.29 -> 27.04 | Improved |
| apply FlowDec instead of Dec-DPM | NaVid · RxR-CE (avg of 6 corruptions) | SR (%): 25.01 -> 33.63 | Improved |
| apply blind denoiser SCUNet to corrupted inputs | NaVid · R2R-CE (avg of 6 corruptions) | SR (%): 21.58 -> 14.68 | Hurt |
| use FlowDec instead of Dec-DPM for decorruption latency | NaVid · Inference time per step | time (ms): 1411 -> 173 | Improved |
| add FlowDec to NaVid on clean inputs | NaVid · R2R-CE (clean) | SR (%): 41.40 -> 40.89 | No clear effect |
| add FlowDec on real robot | NaVid · Real-world Task1 | success rate: 0.20 -> 0.35 | Improved |
| use FlowDec as plug-in to Uni-NaVid | Uni-NaVid · R2R-CE (avg of 6 corruptions) | SR (%): 30.71 -> 32.49 | Improved |
| replace per-frame decorruption with FlowDec temporal consistency | NaVid · R2R-CE warp error | warp error: 0.25 -> 0.11 | Improved |
| increase Euler solver steps from 5 to 20 (ablation) | FlowDec · R2R-CE (avg of 6 corruptions) | SPL (%): 22.88 -> 23.62 | Improved |
Closest papers by meaning
- DecoVLN: Decoupling Observation, Reasoning, and Correction for Vision-and-Language Navigation (CVPR 2026 · similarity 0.89)
- M$^3$E: Continual Vision-and-Language Navigation via Mixture of Macro and Micro Experts (ICLR 2026 · similarity 0.88)
- STEP-Nav: Spatial-Temporal Efficient Visual Token Pruning for Vision-and-Language Navigation with Large Language Models (AAAI 2026 · similarity 0.88)
- Ground Slow, Move Fast: A Dual-System Foundation Model for Generalizable Vision-Language Navigation (ICLR 2026 · similarity 0.87)
- monoVLN: Bridging the Observation Gap between Monocular and Panoramic Vision and Language Navigation (ICCV 2025 · similarity 0.87)
- Adaptive Residual-Update Steering for Low-Overhead Hallucination Mitigation in Large Vision-Language Models (ICML 2026 · 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.
