Work in progress
Elena Tappia, Emilio Moretti, Uta Mohring, Ivo Adan
2023
2023
Cite
Cite
APA
Click to copy
Tappia, E., Moretti, E., Mohring, U., & Adan, I. (2023). Data-driven dynamic scheduling of autonomous mobile robots for part-feeding to assembly lines: Benefits of a digital twin-based approach.
Chicago/Turabian
Click to copy
Tappia, Elena, Emilio Moretti, Uta Mohring, and Ivo Adan. “Data-Driven Dynamic Scheduling of Autonomous Mobile Robots for Part-Feeding to Assembly Lines: Benefits of a Digital Twin-Based Approach,” 2023.
MLA
Click to copy
Tappia, Elena, et al. Data-Driven Dynamic Scheduling of Autonomous Mobile Robots for Part-Feeding to Assembly Lines: Benefits of a Digital Twin-Based Approach. 2023.
BibTeX Click to copy
@unpublished{elena2023a,
title = {Data-driven dynamic scheduling of autonomous mobile robots for part-feeding to assembly lines: Benefits of a digital twin-based approach},
year = {2023},
author = {Tappia, Elena and Moretti, Emilio and Mohring, Uta and Adan, Ivo}
}
Autonomous mobile robots are often used for part-feeding to mixed-model assembly lines due to their high flexibility. Scheduling a fleet of mobile robots in this environment is a complex and dynamic decision problem, which is affected by multiple uncertainties, such as stochastic robot failures and stochastic processing times. Although there is massive real-time manufacturing data available, data-driven and digital twin-based control mechanisms for part-feeding systems have been barely studied in the existing literature.
In this paper, we introduce a digital twin of part-feeding systems to mixed-model assembly lines using a fleet of mobile robots to replenish the buffers at the assembly stations. The digital twin incorporates an agent-based simulation model to mirror the physical system and an integrated two-phase task assignment procedure to dynamically schedule the robots based on real-time system data. We compare multiple data-driven task assignment strategies, varying regarding the amount of incorporated real-time system data. In computational experiments, we find that digital twin-based decision-making for task assignment improves the synchronisation between assembly and material supply as it allows shorter replenishment lead times and lower line inventory levels. Moreover, synchronisation significantly improves the effectiveness of the part-feeding system by reducing the station idle time.
In this paper, we introduce a digital twin of part-feeding systems to mixed-model assembly lines using a fleet of mobile robots to replenish the buffers at the assembly stations. The digital twin incorporates an agent-based simulation model to mirror the physical system and an integrated two-phase task assignment procedure to dynamically schedule the robots based on real-time system data. We compare multiple data-driven task assignment strategies, varying regarding the amount of incorporated real-time system data. In computational experiments, we find that digital twin-based decision-making for task assignment improves the synchronisation between assembly and material supply as it allows shorter replenishment lead times and lower line inventory levels. Moreover, synchronisation significantly improves the effectiveness of the part-feeding system by reducing the station idle time.
