MSNav: Zero-Shot Vision-and-Language Navigation with Dynamic Memory and LLM Spatial Reasoning

TL;DR

MSNav is a robust framework for vision-and-language navigation that combines memory, spatial reasoning, and LLM-based decision-making, significantly improving navigation success rates in complex environments.

Contribution

It introduces a novel integrated architecture with dynamic memory, spatial reasoning, and LLM path planning, along with a new dataset and fine-tuned LLM for enhanced spatial inference.

Findings

MSNav achieves state-of-the-art results on R2R and REVERIE datasets.

Qwen-Spatial outperforms other LLMs in object list extraction.

Memory module improves long-range exploration.

Abstract

Vision-and-Language Navigation (VLN) requires an agent to interpret natural language instructions and navigate complex environments. Current approaches often adopt a "black-box" paradigm, where a single Large Language Model (LLM) makes end-to-end decisions. However, it is plagued by critical vulnerabilities, including poor spatial reasoning, weak cross-modal grounding, and memory overload in long-horizon tasks. To systematically address these issues, we propose Memory Spatial Navigation(MSNav), a framework that fuses three modules into a synergistic architecture, which transforms fragile inference into a robust, integrated intelligence. MSNav integrates three modules: Memory Module, a dynamic map memory module that tackles memory overload through selective node pruning, enhancing long-range exploration; Spatial Module, a module for spatial reasoning and object relationship inference that improves endpoint recognition; and Decision Module, a module using LLM-based path planning to execute robust actions. Powering Spatial Module, we also introduce an Instruction-Object-Space (I-O-S) dataset and fine-tune the Qwen3-4B model into Qwen-Spatial (Qwen-Sp), which outperforms leading commercial LLMs in object list extraction, achieving higher F1 and NDCG scores on the I-O-S test set. Extensive experiments on the Room-to-Room (R2R) and REVERIE datasets demonstrate MSNav's state-of-the-art performance with significant improvements in Success Rate (SR) and Success weighted by Path Length (SPL).