This study tackles the multi-objective scheduling problem for hybrid flow shops, focusing on optimizing makespan and tardiness while adapting to user-defined priorities. We propose a reinforcement learning based solution with a time-dynamic environment that enables realtime decision-making. The relative importance of objectives is directly integrated into the observation space. Evaluated across four setups of varying complexity, the approach shows promising results for simpler setups but faces challenges with scalability, fixed time steps, and optimizing correlated objectives as complexity grows. While rudimentary, this work identifies critical areas for improvement and establishes a foundation for more effective future solutions.     «

This study tackles the multi-objective scheduling problem for hybrid flow shops, focusing on optimizing makespan and tardiness while adapting to user-defined priorities. We propose a reinforcement learning based solution with a time-dynamic environment that enables realtime decision-making. The relative importance of objectives is directly integrated into the observation space. Evaluated across four setups of varying complexity, the approach shows promising results for simpler setups but faces c...     »