During several decades, the research at Waseda University has been focused on developing anthropomorphic robots capable performing musical instruments. As a result of our research efforts, the Waseda Flutist Robot WF-4RIV and the Waseda Saxophonist Robot WAS-1 have been designed to reproduce the human player performance. As a long-term goal, we are proposing to enable the interaction between musical performance robots as well as with human players. In general the communication of humans within a band is a special case of conventional human social behavior. Rhythm, harmony and timbre of the music played represent the emotional states of the musicians. So the development of an artificial entity that participates in such an interaction may contribute to the better understanding of some of the mechanisms that enable the communication of humans in musical terms. Therefore, we are not considering a musical performance robot (MPR) just as a mere sophisticated MIDI instrument. Instead, its human-like design and the integration of perceptual capabilities may enable to act on its own autonomous initiative based on models which consider its own physical constrains. In this paper, we present an overview of our research approaches towards enabling the interaction between musical performance robots as well as with musicians.