City Navigation in the Wild: Exploring Emergent Navigation from Web-Scale Knowledge in MLLMs

This article, sourced from ArXiv, focuses on the application of Multimodal Large Language Models (MLLMs) for city navigation. It investigates how these models can leverage web-scale knowledge to achieve emergent navigation capabilities. The research likely explores the challenges and potential of using MLLMs for real-world navigation tasks, potentially including aspects like route planning, landmark recognition, and adapting to dynamic environments.