AI and Dependency Graph-Based Digital Twins for Context-Aware 6G Robotic Networks

6G robotic networks demand highly context-aware and intelligent management to meet extreme performance requirements. These requirements become even more challenging with real-time control and autonomous decision-making scenarios of 6G robotic applications. However, traditional methods fail to provide context-awareness and enhanced cognitive capabilities to meet these requirements in a robotic network. Therefore, in this paper, we introduce a novel dependency graph-based Digital Twin (DT) framework to provide context-awareness and cognitive decision-making capabilities in robotic applications. In this framework, we design a Data Distribution Service (DDS)-based DT layer and a context-aware mediator layer, comprising a Global What-If Engine and an AutoML-based Decision Unit. The What-If Engine constructs four types of dependency graphs based on temporal, data, control, and performance-related information to represent relationships between robots, sensor inputs, and control triggers. Moreover, the AutoML-based Decision Unit selects the most successful algorithm based on the prediction error and scenario conditions. Experimental results demonstrate that our proposed framework provides real-time DT synchronisation under even high-density twin scenarios without computationally overloading the system. The proposed framework also achieves effective learning-based decision-making for regression and classification tasks in 6G robotic applications.