Performance Categorization for Personalized Learning in Vocational Training Simulators

Simulators have been increasingly used at many industries to provide skill training to its trainees or employees. A haptics based re-bar bending simulator is being used at leading construction training centers in India. The simulator collects the real-time training data and uses it to generate accurate assessment reports similar to the manually generated reports. To enhance the learning environment, the simulator should learn to recognize the differences between the trainee performances and should be capable to model the successive stages of skill learning. In this paper, we discuss how the time-series segmentation technique can be applied to extract process relevant data from the training history collected by the re-bar bending simulator. We applied machine learning methods, on this transformed data to model the varying levels of expertise. In future, these models can be utilized to enhance the simulator to provide personalized feedback and learning experiences.