Abstract: Traditional disembodied intelligent command and control for autonomous unmanned systems relies on absolute coordinates and static maps, resulting in decoupled cognition and action. Such a paradigm faces severe latency and rigidity challenges, struggling to meet the resilience demands of high-intensity intelligent wars. The embodied cognition is fused with global workspace theory to propose an embodied intelligent spatio-temporal topological cognition model centred on the agent’s own body as a measurer. This model firstly establishes an ontological capability measurement framework encompassing hardware variability, software adaptability, and energy affordance, enabling dynamic self-state perception for the intelligent agents. Subsequently, a task-oriented seven-dimensional spatio-temporal topological cognition mechanism is proposed, achieving a fundamental shift in mission feasibility assessment from static, “God’s-eye view” presets to dynamic, embodied-coupled solutions. By incorporating a global workspace mechanism, the model forms a closed-loop “perception-evaluation-decision-reconstruction” process. This drives the dynamic evolution of spatio-temporal topology of the system amidst disturbances, thereby providing a clear theoretical foundation and an engineering-ready implementation pathway for constructing resilient command and control systems with anti-destruction, adaptive, and self-reconfiguring capabilities.