The Music & Entertainment Technology Laboratory (MET-Lab) is interested in
developing robots that can participate with humans in musical ensembles. Such
robots would be useful both as performance devices and as research platforms to
study music and musical performance. In order to enable our platforms to
determine an appropriate response to a musical signal, we have constructed Music
Information Retrieval (MIR) algorithms that analyze musical audio and identify
high level features, such as score location, beat location, and mood. These
algorithms take into account the particular constraints inherent in using a
robot, such as the robot's ego noise. We are also developing motion algorithms
that are parameterized by the musical features so that the robot can move in a
manner congruent with the audio. Our current work is being performed on Hubo, an
adult-sized humanoid robot with the ability to move in a smooth and fluid
manner.