A Cognitive Digital Twin for Intention Anticipation in Human-Aware AI

Intelligent systems are usually not able to adapt to humans but also show difficulties in adapting to varying task sequences, dynamic situations, and context information. This paper introduces a cognitive approach for a digital twin of the human that complements the digital twin of a technical system (e.g., robot) towards a holistic human–machine system. This cognitive digital twin anticipates dynamic situation understanding of the human in-the-loop, uses flexible task knowledge to simulate the human in the real world and integrates further context information. What internal models and mental representations are most relevant for a cognitive digital twin in dynamic interactive situations? The realization of the concept of a theory driven digital approach is further explained in two examples within different application domains. Potentials and research directions for interdisciplinary work are discussed.