Hierarchical DRL Approach for Resource-Constrained Project Scheduling Problem: A Critical Path Optimization Approach

Authors

Vatsal Maru, University of Texas at DallasFollow
Gaurav Shekhar, University of Texas at DallasFollow

Abstract

This paper presents a novel hierarchical deep reinforcement learning approach for resource-constrained project scheduling that explicitly prioritizes critical path optimization. Our methodology integrates graph neural networks with proximal policy optimization using a three-tier reward structure designed to align with established project management principles. Experimental results on PSPLIB benchmarks demonstrate a 14% makespan improvement compared to traditional heuristics on J30 instances with 100% constraint satisfaction. The approach provides both algorithmic advancement and practical insights for intelligent project management systems.

Recommended Citation

Maru, Vatsal and Shekhar, Gaurav, "Hierarchical DRL Approach for Resource-Constrained Project Scheduling Problem: A Critical Path Optimization Approach" (2025). International Research Workshop on IT Project Management 2025. 9.
https://aisel.aisnet.org/irwitpm2025/9