論文ID: TETSU-2024-063
A slab yard within a steel factory comprises multiple Last-In, First-Out (LIFO) buffers, typically managed by a crane operator in a dynamic environment. The efficacy of decision-making in controlling the slab yard hinges on the operator’s cognitive grasp of the task. Therefore, it is crucial to evaluate various cognitive frameworks to enhance and stabilize performance, bolster resilience, and adequately support the operator. This study presents a framework integrating feature variables that convey information about the due dates of slabs, intended for use by and provision to the operator. Subsequently, it employs a combination of behavioural and computational methodologies to assess this framework, utilizing a serious game model of the task for testing purposes. The findings confirm that the effective representation for conveying due date information depends on yard congestion, with the provision of such information potentially backfiring when the yard is crowded. This observation holds true for both computational experiments using a reinforcement learning agent and behavioural experiments using human subjects. Moreover, the consistency of results across both experiments suggests that a reinforcement learning agent could be valuable for formulating plausible hypotheses regarding the suitable cognitive framework for individuals tasked with this responsibility.
