Reinforcement learning in the context of container terminals

Abstract Reinforcement learning (RL) has emerged as a promising paradigm in optimizing various problems in logistics domains, with container terminals standing out as a key area where RL applications hold significant potential for enhancing planning and operations. This comprehensive literature review explores the expansive landscape of RL applications within container terminals, aiming to systematize and analyze the evolution, diverse applications, research insights, and implications of RL in this crucial domain. The study systematically categorizes the utilization of RL across different aspects of container terminal operations, highlighting its varied applications. Moreover, the review critically examines challenges and identifies research gaps in the context of maritime logistics, offering insights for future research directions. Across the 71 examined papers, Q-learning emerges as the predominant RL approach. Notably, scheduling has become the most recurrent problem addressed through RL methodologies, highlighting the pivotal role RL plays in optimizing scheduling processes within container terminals. This extended review provides a nuanced understanding of the contributions of RL, emphasizing its versatility and underscoring its potential to improve container terminal logistics.