Creative Robotics

Course description
The module focuses on developing models of human intelligence in the context of creative activities, where such models can be implemented, tested, and further refined via computational and robotic (embodied) systems. By creative activities, we understand those displayed in the visual arts, music performance, dance, crafts and related human activities. The body, its use in a creative activity, such as drawing, provides constraints which are not naturally taken into account with a traditional computational approach. The body and its uses will impact on the engineering of design system. It will also prescribe what movements are likely and desirable and those which are not. It impacts on the strategies put in place: e.g. one draws using a limb, with a hand, which may get in the way of what is visible.
The course will take a multidisciplinary approach. We will consider how robotics systems can be built by combining a deeper understanding from: the psychology of visual perception, human movement science (such as found in graphonomics), the psychology of visual art, AI for creative systems and artistic support, recent progress with Neural networks (such as Deep Learning approaches used to isolate and transfer painting styles). Quality evaluation by robotics systems will be introduced: how can a robot be able to measure the quality of an artefact either as it is produced or once completed; for example, what is the influence of movements used to create a drawing or painting. In the course, we will consider a number of philosophy of AI questions, such as: How to make a robotic system a useful collaborator to the human artist? What other applications, e.g. in training and education are possible? What is the impact of movement mimicry for the social acceptance of robots? The state of the art at the intersection of AI, robotics, creative and artistic applications will be presented.

On completion of the course, the student should be able to:

  • understand the importance of movement in artistic or creative practices and applications
  • be able to describe some current state-of-the-art embodied creative systems and applications
  • be able to design a system architecture for a simplified robot arm with sensors ( towards creative applications (e.g. drawing)
  • understand how recent machine learning can be used for creative robotics; e.g. compliant motor control, learning from examples, deep learning for style transfer
  • be aware of what advances in technologies will help make robots more able to collaborate with humans in creative practices (and why) e.g. in soft robotics, human movement science, perception, software engineering, haptics

Academic Instructor
Frederic Fol Leymarie|Goldsmiths University