Decisão consensual em sistemas multirrobôs

Abstract

Multi-robot systems (MRS) are generally used to increase the efficiency of smart factories applications. A MRS is a system where two or more robots work in a coordinated or cooperative way with the same goal. This cooperation between multi robots aims to complete multiple tasks in parallel, unattainable for a single robot. In addition, this collective behavior improves the performance, capability, and robustness of task completion. An MRS also presents flexibility since the group can be increased if more tasks need to be attended to. Another advantage of working with MRS is that there is no single point of failure in the system. Thus, if one robot fails, another MRS robot can take over its task. However, it is necessary to draw up a plan in case failures occur in some robots to be replaced by another in acceptable conditions. In this thesis, this behavior is called fault resilience, that is, the ability of a system to recover and continue its operations, despite system failures. In this sense, this thesis aims to generate solutions to manage the collective behavior of multi-robot systems in smart factories to obtain resilience to failures. Thus, a consensus-based, fault-resilient mechanism called Consensual Fault-Resilient Behavior (CFRB) is presented. CFRB is methodized in three hierarchical plans: imposition, negotiating, and consensus. The fault restoration is a consequence of collective behavior managed through the ternary decision in these plans. The approach contributes to production improvements in smart factories, where autonomous multi-robots are employed to improve reliability and robustness in the occurrence of failures. The evaluation of CFRB is performed through experimentation in small-scale warehouse logistics, referred to as Augmented Reality to Enhanced Experimentation in Smart Warehouses (ARENA). Experiments were conducted to show the benefits of the proposed solutions as well as quantitative analysis. In addition, the CFRB was compared with other current approaches, and the results are presented, demonstrating its efficiency.