Embodied intelligence for the intelligent construction of hydraulic engineering： current status， challenges， and prospects

Embodied intelligence， which emphasizes the continuous interaction of physical agents with environments to achieve closed loops of perception， cognition， and decision-making， represents a significant direction in the development of the next generation of artificial intelligence. Its application in hydraulic engineering has fostered a new paradigm characterized by the deep integration of artificial intelligence （AI） models， intelligent construction equipment， and system platforms with virtual and real environments. Through continuous interactive learning and autonomous evolution， this paradigm enables the collaborative optimization of multiple objectives， including progress， quality， safety， and sustainability in hydraulic engineering construction. Using a knowledge graph approach， this paper systematically reviews the development trends of AI-enabled engineering construction and intelligent hydraulic engineering. From the perspective of embodied intelligence， it comprehensively elaborates on the connotation， system architecture， core elements， and characteristics of intelligent hydraulic construction， focusing on typical scenarios such as intelligent unmanned filling， intelligent unmanned compaction， and intelligent construction simulation. Currently， hydraulic engineering construction is evolving from digitization， informatization， and localized intelligence toward the integration of virtual and real systems and systemic collaboration. However， research remains predominantly focused on areas such as intelligent perception， state monitoring， automated process control， and localized decision optimization. Significant challenges persist regarding the core functions of embodied intelligence， such as deep understanding in complex environments， autonomous decision-making， and continuous evolution， as well as in aspects like group collaboration and trustworthy evolution. This paper provides a new theoretical framework for the future development of intelligent hydraulic engineering construction.

具身智能（Embodied Intelligence）强调物理智能体与环境的持续交互，以实现感知、认知与决策的闭环，是下一代人工智能发展的重要方向。其在水利工程领域的应用，催生了以人工智能（AI）模型、智能施工装备、系统平台与虚实环境深度融合为核心特征的全新范式。通过持续交互学习与自主进化，该范式可实现水利工程施工中进度、质量、安全与可持续性等多目标的协同优化。本文采用知识图谱方法，系统梳理了AI赋能工程施工与智能水利工程的发展趋势。从具身智能视角出发，本文全面阐述了智能水利施工的内涵、系统架构、核心要素与特征，重点聚焦智能无人填筑、智能无人压实、智能施工仿真等典型应用场景。当前，水利工程施工正从数字化、信息化、局部智能化向虚实系统融合与全系统协同方向演进。但当前研究仍主要集中于智能感知、状态监测、自动化流程控制与局部决策优化等领域。而就具身智能的核心功能而言，复杂环境深度理解、自主决策与持续进化，以及群体协作、可信进化等方面仍存在诸多重大挑战。本文为智能水利工程施工的未来发展提供了全新的理论框架。