ROS-based Digital Twin Framework for Operator Training in Virtual Reality

Virtual Reality (VR) can create a cost-effective and safe environment for operator training. VR enables the visualization of additional information that is difficult to present on real equipment. However, the impact of additional visual cues on the learning process and transfer of learning outcomes remains insufficiently studied. We offer a Digital Twin based reference architecture and a developed virtual environment to explore the transfer of learning outcomes and the impact of additional visual cues. We designed a series of three object movement tasks to evaluate the effectiveness of robot operator training in three groups: "Real" – on a real robot, "VR" - in VR, "VR+" - in VR with additional visual cues. Thirty participants performed three runs to learn how to control the robot in groups and one control run on a physical robot. For each task, we recorded the following objective performance metrics: time taken, the number of button inputs, the number of joystick inputs, and the number of operator mistakes (cube drops). Although the learning trajectory differed between groups, a comparison of the control run showed no significant difference across all metrics. Additional visual cues contributed to improved performance during training, but they also created a dependency that limited transfer of the learning outcomes when such cues were removed. Subjective ratings were recorded via questionnaires: the NASA TLX workload questionnaire, the User Experience Questionnaire, and the Presence Questionnaire for participants who completed the virtual training. The analysis did not reveal significant differences between the groups for subjective measures.