Teaching machine learning in robotics interactively: the case of reinforcement learning with Lego®Mindstorms
Ángel Martínez-Tenor, Ana Cruz‐Martín, Juan‐Antonio Fernández‐Madrigal · Interactive Learning Environments · 2018
Preparing students for dealing with a world more and more densely populated with physical machines that possess learning capabilities, e.g. intelligent robots, is of the utmost importance in engineering. In this paper, we describe and analyse a design of interactive sessions devoted to the application of some machine learning (ML) methods within a master degree subject named “Cognitive Robotics”, in particular, reinforcement learning (RL), a technique that allows the machine to autonomously learn decision-making in a physical environment. The paper contains a complete depiction of the interactive teaching sessions, implemented with reasonable cost and resources thanks to a suitable mixture of the constructivist and instructivist paradigms. It also gathers the experiences of students and teachers through both qualitative and quantitatively indicators.