2026 ASEE Annual Conference & Exposition

As collaborative robots (cobots) become increasingly prominent in advanced manufacturing, en-gineering education must adapt to equip students with the skills needed to safely and efficiently operate and program these systems. However, traditional robotics instruction often involves steep learning curves, complex programming languages, and limited access to real-world hardware. To address these challenges, this research presents a narrative-driven digital twin framework, which leverages Generative AI and Unreal Engine-based simulation, to enhance collaborative robotics education through immersive, intuitive, and safety-aware virtual training environments. Our pro-posed framework integrates large language models with a 3D digital twin of the UR3e collabora-tive robot, built using Unreal Engine. Students can input narrative-based instructions, which are interpreted by OpenAI’s GPT API to generate high-level command sequences. These commands are then simulated within the digital twin environment, allowing students to visualize and validate cobot behaviors without physical hardware. This lowers the barrier to entry, enabling learners with limited programming backgrounds to engage meaningfully with robotics tasks.
During the current summer research phase, two undergraduate students participated in the initial implementation and testing of the platform. Key accomplishments include (1) completion of the UR3e robot digital twin in Unreal Engine, and (2) partial integration of the OpenAI API to gen-erate textual commands from natural language prompts. While integration with real-time sensor data (Intel RealSense D435i) and edge processing hardware (Jetson Orin Nano) remains under development, the simulation environment is fully functional and serves as a foundation for future extensions. The framework is designed to be scalable and adaptive. Future integration of algo-rithms such as Deep Reinforcement Learning (DRL), Multi-Objective Evolutionary Algorithms (MOEAs), and Physics-Informed Neural Networks (PINNs) will support motion optimization and safety assurance. Additionally, Explainable AI (XAI) tools will be incorporated to provide transparency in AI decision-making, which is essential for student trust and comprehension in safety-critical robotics systems.
Preliminary educational insights from the project indicate strong student engagement and rapid skill acquisition when using narrative-driven AI tools in conjunction with immersive simulation. Students reported increased confidence in understanding robotic kinematics and spatial reasoning, as well as improved motivation to explore advanced AI techniques. This work contributes to the growing body of educational technology research by introducing a novel paradigm that fuses Generative AI, digital twins, and simulation-based learning. By abstracting the technical com-plexities of cobot programming and integrating real-time visual feedback, this platform offers a transformative approach to teaching robotics in undergraduate curricula.
The paper will present the architectural design, development workflow, and partial implementation results, along with proposed strategies for full integration and classroom deployment. Future plans include connecting real-world sensor streams for hybrid simulation and developing structured modules for active classroom use, assessment, and collaborative learning. This study demonstrates the feasibility and potential of leveraging modern AI and simulation tools to create accessible, safe, and engaging educational platforms for collaborative robotics training.