Production planning and simulation in mixed reality for human work performance variations

Abstract

Manual material handling is still common in most factories nowadays. A production system involving manual tasks can experience significant uncertainty due to variations in individual human actions. Previous research has not fully addressed production simulation and planning in this context. This study proposes a systematic simulation approach that combines Mixed Reality (MR) with offline system simulation to characterize the uncertainty introduced by individuals in manual tasks. An MR-based simulation tool allows a manager and an operator, both using HoloLens, simultaneously collaborate in a facility constructed by overlaying virtual objects in the real environment. An initial simulation experiment is conducted to identify human attributes that could influence work performance of the facility. Supported by FlexSim, statistical analyses are used to incrementally estimate system parameters related to manual tasks with increasing complexity in order to meet a predefined target of the system after a new operator has replaced the current one. The simulation tool also incorporates decision-making rules derived through K-Mode clustering to instantly prompt the operator to adjust their work performance during simulation. A case study shows how the proposed approach assists to balance the part supply and demand through manual material transportation using a cart trolley in a logistics facility. This work validates the feasibility of MR as a new paradigm for production simulation involving humans, overcoming the deficiencies of traditional simulation software.