A self-evolving, training-free VLN agent with hierarchical memory, RAG plus chain-of-thought reasoning, and reflection reports state-of-the-art success rates on R2R and REVERIE.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CV 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
SE-VLN: A Self-Evolving Vision-Language Navigation Framework Based on Multimodal Large Language Models
A self-evolving, training-free VLN agent with hierarchical memory, RAG plus chain-of-thought reasoning, and reflection reports state-of-the-art success rates on R2R and REVERIE.