PPSTOW: An End-to-End Deep Reinforcement Learning Model for Master Stowage Planning on Container Vessels

Efficient supply chains are vital for both the worldwide economy and environmental sustainability. Container shipping plays a key role in this, known for being an eco-friendly mode of transport. Liner shipping companies are actively working to improve operational efficiency through stowage planning. Due to many combinatorial aspects, some of which are NP-hard, stowage planning is a challenging problem in its representative form. Even though stowage planning can be decomposed into master and slot planning, the subproblems remain challenging. As a result, we are searching for scalable algorithms to solve the stowage planning problem.

In this work, we propose Proximal Policy optimization for master STOWage planning (PPSTOW), a deep reinforcement learning approach to address master planning with focus on global problem objectives and constraints. The experiments show the effectiveness of PPSTOW, as the framework efficiently finds near-optimal solutions for simulated problem instances with realistic vessel sizes and practical planning horizons. In the future, we aim to refine the representativeness of our approach by integrating revenue management, as well as local problem objectives and constraints.