Per-Shipment Multi-Agent Reinforcement Learning for Intermodal Freight Routing Under Hurricane Disruption

2026-08-07

AI Summary

This study addresses the challenge of simultaneous multi-channel disruptions in intermodal freight networks caused by extreme weather events such as hurricanes. The authors propose a resilience-oriented routing approach that models shipment-level routing decisions as a decentralized partially observable Markov decision process (Dec-POMDP) and employs a centralized training with decentralized execution (CTDE) framework. For the first time in this domain, multi-agent reinforcement learning is applied at the shipment granularity, implementing both Independent PPO (IPPO) and MAPPO policies. Experimental results demonstrate that IPPO significantly outperforms MAPPO in throughput (+12.7%), delivery rate, and resilience index under high load (+6.4%). MAPPO suffers from severe performance degradation due to distributional shift between training and evaluation queue dynamics, remaining inferior even after retraining. The findings highlight IPPO’s superiority in capacity-constrained scenarios and establish a new paradigm for optimizing intermodal networks under disruption.

Abstract

Intermodal freight networks face growing disruption risk from climate extremes that degrade multiple corridors simultaneously. To address this, we formulate freight routing as a Dec-POMDP with per-shipment action granularity and train Independent PPO (IPPO) under Centralized Training with Decentralized Execution, comparing against two heuristic baselines with privileged state access on a 15-hub network under hurricane disruption. Across 30 matched episodes, no single policy dominates: IPPO achieves the highest throughput ($+12.7\%$) and delivery rate while a capacity-aware heuristic leads on Resilience Index (RI) and delay. Under demand surge (2.9:1 capacity ratio), IPPO's RI advantage grows to $+6.4\%$, suggesting learned routing is most valuable when capacity is scarce. A Multi-Agent PPO (MAPPO) variant collapses under train-eval queue mismatch ($\mathrm{RI} = 0.811$); retraining recovers RI to $1.018$ but IPPO still leads on throughput, pointing to residual limitations in centralized critics under per-shipment dispatch.

Problem

• Intermodal Freight Routing
• Hurricane Disruption
• Multi-Agent Reinforcement Learning
• Dec-POMDP
• Resilience

Innovation

• Multi-Agent Reinforcement Learning
• Per-Shipment Routing
• Dec-POMDP
• Independent PPO
• Intermodal Freight Resilience