StreamVLN: Streaming Vision-and-Language Navigation via SlowFast Context Modeling

TL;DR

StreamVLN introduces a hybrid slow-fast context modeling framework for real-time vision-and-language navigation, enabling efficient, coherent multi-turn dialogue understanding over long visual streams with low latency.

Contribution

It proposes a novel slow-fast context modeling strategy that balances visual understanding, long-term context, and computational efficiency in streaming VLN tasks.

Findings

Achieves state-of-the-art performance on VLN-CE benchmarks.

Supports long video streams with bounded context and low inference cost.

Demonstrates robust and efficient real-world deployment capabilities.

Abstract

Vision-and-Language Navigation (VLN) in real-world settings requires agents to process continuous visual streams and generate actions with low latency grounded in language instructions. While Video-based Large Language Models (Video-LLMs) have driven recent progress, current VLN methods based on Video-LLM often face trade-offs among fine-grained visual understanding, long-term context modeling and computational efficiency. We introduce StreamVLN, a streaming VLN framework that employs a hybrid slow-fast context modeling strategy to support multi-modal reasoning over interleaved vision, language and action inputs. The fast-streaming dialogue context facilitates responsive action generation through a sliding-window of active dialogues, while the slow-updating memory context compresses historical visual states using a 3D-aware token pruning strategy. With this slow-fast design, StreamVLN achieves coherent multi-turn dialogue through efficient KV cache reuse, supporting long video streams with bounded context size and inference cost. Experiments on VLN-CE benchmarks demonstrate state-of-the-art performance with stable low latency, ensuring robustness and efficiency in real-world deployment. The project page is: \href{https://streamvln.github.io/}{https://streamvln.github.io/}.