ECAI 2025 · 2025
EcoFair-CH-MARL: Scalable Constrained Hierarchical Multi-Agent RL with Real-Time Emission Budgets and Fairness Guarantees
Saad Alqithami
Frontiers in Artificial Intelligence and Applications, Volume 413
This paper introduces EcoFair-CH-MARL, a constrained hierarchical multi-agent reinforcement learning framework that unifies three innovations: a primal–dual budget layer that provably bounds cumulative emissions under stochastic weather and demand; a fairness-aware reward transformer with dynamically scheduled penalties that enforces max–min cost equity across heterogeneous fleets; and a two-tier policy architecture that decouples strategic routing from real-time vessel control, enabling linear scaling in agent count.
Emissions Reduction
Up to 15% lower emissions
Throughput
12% higher throughput
Fair-Cost Improvement
45% improvement over baselines
Regret Bound
O(√T) for constraint violations
Multi-Agent RLEmission BudgetsFairnessMaritime LogisticsHierarchical Learning
Read the full paper