Behavioral programming, decentralized control, and multiple time scales

Abstract

Behavioral programming is a recently proposed approach for non-intrusive incremental software development. We propose that behavioral programming concepts, such as behavioral decomposition, synchronized execution of independent behaviors, and event blocking, can help in the incremental and natural coding of complex decentralized systems, complementing actor-oriented and agent-oriented approaches. We also contribute to the existing research on behavioral programming a method for coordinating behaviorally-programmed components which, due to different time scales or interaction with the external environment, cannot synchronize and thus cannot employ event blocking. We show that the resulting decentralized system retains many of the advantages present in a purely behavioral, fully synchronized system.