Decentralized task allocation for heterogeneous teams with cooperation constraints

Abstract

This paper presents decentralized methods for allocating heterogeneous tasks to a network of agents with different capabilities, when the rules of engagement dictate various cooperation constraints. The new methods are built upon the consensus-based bundle algorithm (CBBA), and the key extensions to the baseline CBBA are: (a) task decomposition and associated scoring modification to allow for soft-constrained cooperation preference, and (b) a decentralized task elimination protocol to ensure satisfaction of the hard-constrained cooperation requirements. The new extensions are shown to preserve the robust convergence property of the baseline CBBA, and numerical examples on a cooperative track & strike mission verify the performance improvement by these extended capabilities.