StreamVLN: Streaming Vision-and-Language Navigation via SlowFast Context Modeling
This addresses the challenge of efficient and responsive VLN for real-world deployment, though it appears incremental as it builds on existing Video-LLM methods with a novel hybrid approach.
The paper tackled the problem of Vision-and-Language Navigation (VLN) in real-world settings by introducing StreamVLN, a framework that uses a hybrid slow-fast context modeling strategy to balance fine-grained visual understanding, long-term context modeling, and computational efficiency, achieving state-of-the-art performance on VLN-CE benchmarks with stable low latency.
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/}.