This paper explores the current state and future prospects of dyadic Human-Robot Interaction (HRI), focusing on the opportunities and challenges in this evolving field. We examine the role of key technologies, including computer vision, Digital Twins (DTs), and Generative Artificial Intelligence (GenAI), in enhancing the quality and effectiveness of dyadic interactions between humans and robots. Beginning with a contextualisation of dyadic HRI and its foundational concepts, we identify and review the stages of dyadic interaction and how they are all interconnected with the latest advancements in visual perception, and non-visual systems, then highlighting how these technologies can be employed to create more responsive and adaptable robots. Central to our discussion is the integration of DTs and GenAI and the exploration of how they can be leveraged to develop predictive models that improve real-time interactions. We also address the ethical, societal, and regulatory implications of these technological integrations, considering how they might shape the future of human-robot partnerships. Finally, we propose future research directions aimed at advancing the field of dyadic HRI. By providing this analytical review, this paper seeks to offer a concise roadmap for the future of dyadic HRI where technology and human-centred design converge.