TANSA: Task Allocation Using Nomadic Soft Agents for Multirobot Systems

Abstract

The availability of a gamut of ubiquitous and embedded hardware, coupled with rapid advancements in sensing technologies, has thrown open research on a wide range of real-world multirobot applications. Investigating novel mechanisms for the allocation of tasks within such multirobot systems (MRS) forms a prominent area of research. This paper addresses the issue of competing instances of concurrent task allocations occurring within the system, wherein all allocations vie for the same set or subsets of robots. In this paper, we formally describe the problem within an optimization framework and propose an approach for Task Allocation using Nomadic Soft Agents (TANSA) to handle concurrent and multiple task allocation instances in MRS. Nomadic soft agents are autonomous programs (code) that have the ability to migrate and execute within a network. The proposed mechanism uses a novel concept termed mobility resource that aids in regulating multiple instances of task allocations by forcing voluntary and adaptive back-offs on part of the nomadic soft agents. We also discuss a variant of the proposed approach-TANSA* for comparing the significance of adaptive and nonadaptive back-off mechanisms. Extensive experiments have been performed in emulated environments on real networks. The results clearly show that the TANSA-based approach outperforms two other prominent task allocation mechanisms. Further experiments in the real world performed using seven robots fortify the practical viability of this approach.