On November 11, Beijing officially issued the "High-Quality Development Action Plan for the Embodied Intelligent Robot Industry (2026–2028)," which calls for significantly enhancing the capabilities of embodied intelligence foundation models. The plan supports the coordinated layout and integrated development of multiple technical routes—including Vision-Language-Action (VLA) models, World Models (WM), and reinforcement learning—with the aim of driving technological innovation across these major pathways. It also supports breakthroughs in key technologies such as multi-source information fusion, physical space understanding, and human emotion recognition, facilitating a leap from "perception" to "cognition." This will enable "brain"-level cognition to reach a point where systems can understand the constraints of physical laws and proactively recognize non-verbal cues such as human actions and facial expressions. The plan further supports advances in real-time reasoning, cross-embodiment transfer, and self-learning evolution, strengthening "cerebellum"-level capabilities in multi-task generalization, highly reliable execution, rapid operation, and ultra-precision manipulation. In addition, it backs research into swarm collaboration, long-term memory, localized lightweight deployment, and model interpretability—significantly improving the real-world deployment capabilities of embodied intelligence models and driving embodied intelligence foundation models to reach internationally first-class levels overall.