Main research findings
This thesis investigates how robots can learn to play drums, showing that intelligent and creative behaviour can emerge from the robot’s physical design and its interactions. A robot named ZRob was developed with a flexible gripper that can exploit the dynamics of the body and environment for creating emergent rhythmic patterns. By using machine learning, specifically reinforcement learning, ZRob learns rhythmic patterns through interaction. By trial and error, it adapts and improves, sometimes producing surprising and unique rhythms. This study suggests that robots like ZRob can go beyond following set patterns—they can become creative performers in their own way. These findings open possibilities for robots not only as precise, automated musicians but as collaborative partners capable of enhancing human music performances.
Adjudication committee
- Professor Sylvie Gibet, University of Bretagne Sud, France
- Associate Professor Daniel Overholt, Aalborg University, Denmark
- Professor Michael Welzl, University of Oslo, Norway
Supervisors
-
Professor Jim T?rresen, Department of Informatics, University of Oslo, Norway
-
Professor Alexander Refsum Jensenius, University of Oslo, Norway
-
Professor Emeritus Rolf Inge God?y, University of Oslo, Norway
Chair of defence
- Associate Professor Ragnhild Kobro Runde, Department of Informatics, University of Oslo
Recordings
The event was streamed on YouTube. Below are recordings of each section.
Trial lecture
Topic: "AI in music technology to improve human creativity”.
Thesis introduction
First opponent
Second opponent
Contact
Contact information at Department of Informatics: Mozhdeh Sheibani Harat