The Digital Press

This study develops a Deep Reinforcement Learning-based system to assign employees to tasks efficiently in complex, real-world dynamic task assignment problems (DTAPs). The system uses a novel graph-based representation and a reward function aligned with minimizing task cycle time.

Why it matters: Efficient task assignment reduces processing time, boosting operational performance in real-world work environments.

The big picture: The DRL agent generalizes well across different DTAP scenarios, handling large-scale, stochastic task sequences.

Stunning stat: The DRL system outperforms or matches the best baseline methods across five real-world inspired DTAP instances.

Quick takeaway: A graph-based observation/action structure plus a cycle time–minimizing reward enable effective, scalable DTAP decision-making.