As a cutting-edge development in artificial intelligence, embodied intelligence (EAI) demonstrates enormous potential in enriching public cultural service content and innovating service models. However, its current application in libraries is still in the exploratory stage of using "intelligent tools," which limits it from realizing its full potential. Focusing on the human-machine collaborative service model in libraries driven by embodied intelligence, this paper applies multi-modal perception fusion technology, digital twin technology, knowledge graphs, and cloud collaborative computing platforms at the technical level, and provides personnel, institutional, and organizational guarantees at the support layer. Through these measures, an application closed loop of "perception-decision-action-feedback" has been constructed and a human-machine collaborative model covering resource management, user services, activity implementation, and space management has been formed to transform libraries into smart spaces that are capable of autonomous perception, analysis, decision-making, and action.