Man-machine collaborative geospatial reasoning cognitive framework method and system

Abstract

The invention relates to the technical field of artificial intelligence, and discloses a man-machine collaborative geospatial reasoning cognitive framework method and system, and the method comprises the steps: collecting the intention of a user through a multi-modal front end, and dynamically constructing a unified spatial semantic context; decomposing a high-level task into a structured inference chain by utilizing a large language model; a meta-cognitive control module dynamically schedules a model in an AI capability matrix according to the inference chain to execute a task; feeding back a staged execution result to the user; the user and system data are continuously updated to a spatial semantic context cache center, and finally the reasoning process and result are visually presented, and the intervention of the user in the reasoning process is supported. The system comprises an intention understanding module, a task planning module, a meta-cognitive control module, a process adjusting module and a man-machine interaction module. According to the method, a cognition framework integrating context perception, autonomous planning, dynamic execution and closed-loop interaction is constructed, and deep simulation of the whole process of expert-level geographic space cognition is achieved.

Images