Learning to Solve Resource-Constrained Project Scheduling Problems with Duration Uncertainty using Graph Neural Networks
Guillaume Infantes, Stéphanie Roussel, Antoine Jacquet, Emmanuel Benazera
TL;DR
The paper tackles solving the Resource-Constrained Project Scheduling Problem (RCPSP) under duration uncertainty by proposing a proactive offline approach that yields reusable baselines. It models uncertainty as an MDP over AON-flow graphs and learns a scheduling policy with a Graph Neural Network (GNN) driven by Deep Reinforcement Learning, specifically using PPO, within a framework called Wheatley. Key contributions include (i) a formal MDP formulation for RCPSP with uncertainty, (ii) a GNN-based agent with graph rewiring, resource-aware representations, and a virtual pooling node, and (iii) extensive experiments on PSPLib benchmarks showing strong performance and generalization, with Wheatley made publicly available. The work demonstrates that a learned, uncertainty-aware priority rule can outperform traditional Priority Dispatch Rules and scale to unseen datasets, offering a robust, reusable baseline for industrial scheduling under variability.
Abstract
The Resource-Constrained Project Scheduling Problem (RCPSP) is a classical scheduling problem that has received significant attention due to of its numerous applications in industry. However, in practice, task durations are subject to uncertainty that must be considered in order to propose resilient scheduling. In this paper, we address the RCPSP variant with uncertain tasks duration (modeled using known probabilities) and aim to minimize the overall expected project duration. Our objective is to produce a baseline schedule that can be reused multiple times in an industrial setting regardless of the actual duration scenario. We leverage Graph Neural Networks in conjunction with Deep Reinforcement Learning (DRL) to develop an effective policy for task scheduling. This policy operates similarly to a priority dispatch rule and is paired with a Serial Schedule Generation Scheme to produce a schedule. Our empirical evaluation on standard benchmarks demonstrates the approach's superiority in terms of performance and its ability to generalize. The developed framework, Wheatley, is made publicly available online to facilitate further research and reproducibility.