Answers are grounded in this paper and cite their sources.
To support autonomous equipment operations in complex and dynamic scenarios, this paper proposes the concept of an Embodied Intelligent Digital Twin (EIDT) and its implementation guidelines. By adapting and integrating digital twin (DT) technology with a domain-specific foundation model (DSFM), the EIDT links physical entities’ mechanistic characteristics and dynamic states with operational domain knowledge, establishing a task cognition and autonomous operation mechanism jointly driven by mechanism, data, and knowledge. In addition, the EIDT integrates information about Operation Tasks, Operation Performers, Operation Environments, and Operation Targets to support dynamic perception, task understanding, and adaptive decision-making. A unified EIDT representation paradigm comprising a mechanistic model unit (MU), a data model unit (DU), a DSFM, and an interaction interface (I) is proposed, together with an integrated construction solution. Implementation routes are developed for task-driven autonomous operation, adaptation across tasks and scenarios, and continuous evolution across lifecycle stages. Human-robot collaborative assembly of a rotary vector (RV) reducer is used to demonstrate EIDT construction, collaborative work-system formation, autonomous operation, and application extension for collaborative manufacturing tasks. The proposed EIDT provides a theoretical framework and implementation guidelines for integrating foundation models with DTs to support perception, planning, decision-making, and execution in dynamic operational scenarios.
No takes yet. Share an insight, caveat, or question.
Hu et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: