Editorial: Learning-driven optimization for solving scheduling and logistics

Many real-world applications for complex industrial engineering or design problems can be modelled as optimisation problems. These problems often have features such as multi-modality (multiple optimal solutions), dynamics (changing problem conditions over time), discontinuity (sudden jumps or changes in the problem landscape), and nonlinearity (relationships that do not follow a straight line). Prior knowledge of the solved problem and the properties of algorithms are vital for solving complex problems. Therefore, an algorithm should be able to learn from the solving process. Learning-driven methods help agents learn the solution space by taking actions and interacting with the environment. In doing so, agents continuously update their strategies. Reinforcement learning (RL), where an agent learns to make decisions by receiving rewards or penalties based on its actions, and metaheuristic algorithms, which use problem-solving strategies that guide the search process to find near-optimal solutions, are emerging approaches that use advanced computational power to address such challenges. These approaches have been actively investigated and applied, especially in scheduling and logistics operations. New algorithm frameworks and theoretical analysis are new approaches to solving complex problems. Recent research has improved evolutionary algorithms by integrating reinforcement learning, creating reinforcement learning-assisted evolutionary algorithms (RL-EAs). RL-EAs use learned search information to optimise solutions and have succeeded in many domains. However, few studies address theoretical analysis, benchmarks, training methods, or strategy design. RL-EAs still face high computational costs and sparse rewards. Researchers need new methods to improve algorithm performance.

Real-world applications are essential for validating algorithm performance. Learning-driven optimisation algorithms now address more complex optimisation problems that challenge traditional approaches. In scheduling, large data sets optimise routes, taxi dispatch, dynamic bus scheduling, and other mobility services, boosting efficiency. In logistics, advanced data techniques optimise material movement within and between supply chain entities, such as warehouses, factories, distribution centres, and retail shops. These applications remain complex, so no single solution works for all cases. Practitioners must adapt learning-driven and metaheuristic methods to scheduling and logistics.

This Research Topic aims to discuss new and existing Research Topic in these areas, in particular, to explore learning-driven optimisation algorithms for solving scheduling and logistics problems. Six papers from three countries (Zhang et al., Liang et al., Xu et al., Zhang et al., Zhang et al., Japan contribution Kudo et al., Xu et al., Zhang et al., Zhang et al., and Spain contribution Zhang et al. are accepted to published in this Research Topic.

Real-world applications are vital for learning-driven optimisation research. Three research papers are accepted on this Research Topic. A discrete fireworks algorithm (DFWA) is proposed to solve the open capacitated location-routing problem (OCLRP) Zhang et al. A hybrid heuristic method that combines a multi-objective evolutionary algorithm (MoEA-HSS) is proposed to solve the generalised police officer patrolling problem (GPOPP) Kudo et al. A hybrid intelligent algorithm integrating Q-learning is proposed to solve the time-variant berth and quay crane allocation problem Liang et al.

The state-of-the-art advantageous research is collected in three review papers. A systematic and comprehensive review of multi-agent reinforcement learning (MARL) methodologies and their applications to the flexible shop scheduling problem (FSSP) is presented Xu et al. Metaheuristics for multi-objective scheduling problems in Industry 4.0 and 5.0 are introduced Zhang et al., and the enhancement of multi-objective evolutionary algorithms with machine learning for scheduling problems is discussed Zhang et al. With the comprehensive survey on these Research Topic, the future direction of scheduling and logistics can be guided.

We would like to thank the editorial staff and the Journal Manager, Hannah Lee, for their excellent assistance with this Research Topic throughout the process. We are equally grateful to all the authors for their valuable contributions and to the reviewers for their constructive suggestions, which have greatly improved the quality of the papers presented here. We hope that the reader finds the papers collected in this thematic Research Topic interesting and helpful.

Statements