Consensus-based Robust Decentralized Task Assignment for Heterogeneous Robot Networks

This paper considers the problem of decentralized task assignment in a network of heterogeneous robots. We introduce a new algorithm named heterogeneous robots consensus-based allocation (HRCA), which can be viewed as a possible extension of the recently proposed consensus-based bundle algorithm (CBBA) for homogeneous robot networks. The HRCA is based on a two stage decentralized procedure. In the first stage, similarly to CBBA, an initial assignment based on market-based decision strategies and local communication is determined, disregarding possible constraints on the maximum number of tasks assignable to each robot. Constraint violations are handled in the second stage, in which an iterative procedure is used by the robots to redistribute the tasks exceeding their individual capacity with minimal losses in terms of score function. Numerical simulations are used to evaluate the performance of the HRCA in a set of randomly generated scenarios, which include some examples of homogeneous networks to allow a comparison with CBBA.