多模可信交互：从多模态信息融合到人-机器人-数字人三位一体式交互模型
摘要
Trust plays a key role in the intelligent collaboration among humans, robots, and digital humans, as it affects not only the efficiency of their interaction, but also the risks and benefits involved. However, the current research on human-agent trust suffers from an “imbalance" phenomenon; that is, most studies focus on the “human-centered" trust relationship and ignore the trust of intelligent agents (robots and digital humans) towards humans. To fill this gap in the research field of human-robot-digital human trust, we construct a “trinitarian" experience blending and sharing ecology across the virtual and real worlds, based on the use of multimodal signals to establish moderate unidirectional trust in human-robot-digital human interaction. This paper aims to explore the existing research, feasibility, and future development directions of multimodal cues and augmented reality in establishing trustworthy relationships among humans, robots, and digital humans. Finally, this paper envisions the application prospects of this model in social organization and social event situation perception and control, and points out the problems that need to be solved in the future. This paper contributes to the understanding of the role of multimodal cues and augmented reality in human-robot-digital human interaction, and provides ideas and solutions for the realization of the “trinitarian" trend.