Integrating Digital Twins, Mathematical Modelling and Embodied AI to Bridge Simulation–Reality Gaps in Robotics Education: A Qualitative Case Study in South African HEIs

Abstract

Robotic manipulation research in higher education institutions (HEIs) often faces the challenge of bridging the gap between simulation outcomes and real-world performance. This study, grounded in embodied intelligence theory and mathematical modelling, explored how digital twins and embodied AI can be integrated to enhance robotic manipulation in academic contexts. The research objectives were: (1) To investigate how digital twins and embodied artificial intelligence can be integrated to enhance robotic manipulation in higher education contexts. (2) To examine the contribution of mathematical modelling to improving the transfer of skills from simulation environments to physical robotic systems. The study adopted an interpretivist paradigm and a qualitative case study design. Data were collected over a six-month period in 2024 through semi-structured interviews and laboratory observations involving 15 participants, including lecturers, postgraduate students and technical staff from three HEIs with established robotics laboratories. Findings revealed that digital twins, supported by mathematical models, significantly enhance the transfer of manipulation skills from virtual simulations to physical robots. Meanwhile, embodied AI improves adaptability in unstructured environments and fosters effective human– robot collaboration. The study may contribute to policy and practice by recommending broader integration of digital twin platforms, incorporation of embodied AI into robotics curricula, and strengthened institutional collaboration to address resource disparities and promote inclusive robotics education across HEIs.