如何构建通用具身导航大模型？

Core Insights - The article discusses advancements in general navigation models within the field of embodied intelligence, highlighting the transition from task-specific navigation systems to more universal models that can handle a variety of tasks and environments [2][5]. Group 1: Navigation Models - The Uni-NaVid model represents a cross-task navigation framework that aims to enhance the capabilities of navigation systems beyond specific tasks [5][6]. - The NavFoM model is a cross-ontology navigation framework that further expands the application of navigation algorithms to various real-world scenarios, including visual obstacle avoidance and urban micro-mobility [2][5]. Group 2: Applications and Challenges - Current navigation systems struggle with unstructured, dynamic environments and complex tasks requiring language understanding, which traditional systems cannot adequately address [2][5]. - The introduction of navigation large models is seen as a pathway to achieving embodied intelligence by broadening the scope of navigation algorithms from specialized capabilities to general intelligent mobility [2][5]. Group 3: Event Details - A live session featuring Zhang Jiazhao, a PhD student from Peking University, will take place on November 20 from 19:30 to 20:30, focusing on the exploration of general navigation models [5][6]. - The session will cover specific applications of the navigation models, including TrackVLA++, UrbanVLA, and MM-Nav, showcasing their practical implementations [6].