Order-Controlled Production Employing Multi-Agent and Flexible Job-Shop Scheduling on a Physical Simulation Platform

• Alex Luiz de Sousa UDESC
• André Schneider de Oliveira UTFPR

Abstract

This article presents a distributed multi-agent system that replaces a physical factory simulation platform's centralized and fixed production cycle architecture. The physical platform was modified to support flexible order-controlled production, with a flexible job-shop scheduling algorithm to create the production schedules. The new architecture allows agents to control factory production on the simulation platform and cooperate by employing leaderless consensus control. Multi-agent communication is developed with the Robot Operating System framework, and a scheduler is responsible for distributing production schedules to the agents. New manufacturing stations were included in the platform to increase its simulation power and prove the flexibility of the proposed architecture. Experiments showed that the proposed multi-agent architecture could handle many more sequences of operations than the traditional architecture. A final section discusses essential points of the work, followed by conclusions and future objectives.

Keywords: Schedules, Scheduling algorithms, Robot kinematics, Operating systems, Production, System recovery, Production facilities, multi-agent, flexible job-shop, consensus