COMP 150-02 Developmental Robotics

This class serves as an introduction to the interdisciplinary field of Developmental Robotics, which crosses the boundaries between robotics, artificial intelligence, and developmental psychology. The goal of the field is to create autonomous robots that are intelligent and adaptable in the real world rather than in very limited domains, situations and environments. The class will focus on representations and algorithms that enable a robot to continuously learn about its physical or social environment through its own interaction with it. Topics include overview of robotics; robotics cognitive architectures; deep learning for visual and non-visual sensory data; unsupervised, self-supervised, and reinforcement learning in robotics; learning object affordances; and, theories of cognitive development and their applications to robotics. There will be several small homework assignments and one large class project, with the goal of producing work worthy of publication. You will use physics-based robot simulators as well as real robots as part of the final project. By the end of this class you will have an understanding of the current state of the art of the field and will be able to conduct original research within it.