Optimising digital twin laboratories with conversational AIs: enhancing immersive training and simulation through virtual reality

Abstract

Digital twin laboratories, accessible through low-cost, portable virtual reality (VR) headsets, have become a powerful tool in chemical education and research collaboration. These immersive digital environments replicate physical laboratories, offering unique platforms for planning experiments, conducting virtual lab tours, and training on specialist equipment. In this paper, we present the development of Lab427 VR, a digital twin model of our laboratory designed to be a novel platform for global collaborative research with immersive training. A significant advancement in our approach to the potential of digital twins such as our laboratory is the integration of conversational artificial intelligence (AI) avatars, which address operational gaps in current digital twin systems. We designed and trained three specialised AI avatars to perform key laboratory functions, achieving up to 95% accuracy in their responses, assessed using evaluation metrics such as human evaluation, set-based F1 scoring, and BERTScore. Our findings highlight the potential of combining digital twin technology with AI-driven solutions to enhance laboratory collaboration and training, demonstrating the future potential of smart, interactive connected laboratory environments.