Design and Implementation of a Mixed-Reality Robotic Training System to Support Multiple Learning Models

This thesis aimed to develop a mixed reality-based robotic platform that could support user’s learning, stressing out the differences between traditional learning methods and the learning by teaching paradigm, where the learner reinforces his or her own knowledge by giving lessons to a teachable agent, role in this case played by the Cozmo robot. First, a depth research was conducted to understand the current state of this field of study and the related technology, searching for possible lacks with the aim to take a step forward in the tackled domain. Then, an application was designed and implemented in Unity including a specific mode for each learning paradigm, whose topic was based on the Thèvenin Theorem from the electrical domain, aside from a tutorial to master the user interface. The robot acted autonomously during the traditional mode, while in the second mode it was manually controlled following the Wizard of Oz approach, aiming for a more realistic human-robot interaction. For fairness purposes, in both modes the robot was endowed with identical movement and action capabilities. The mixed reality environment was achieved converting a projection into a touch screen applying a background subtraction algorithm to images captured from a depth camera sensor. Both the robot and the sensor were controlled using the Python language. The devised system was evaluated by several participants, whose experience was collected via a standard questionnaire with the objective to possibly validate the advantage of the studied methods under certain conditions. In conclusion, the present thesis managed to design and implement a mixed reality robotic training system able to support two different learning models. The main concern during the development of the system was its usability, as the goal was to implement an intuitive interface that, after a short learning period to master it, could be capable to let the user focus on the learning objectives rather than on technology, a goal which was achieved to a certain extent.