Risk-Conditioned Distributional Hierarchical Deep Q-Network for Multi-Depot Vehicle Routing under Time-Dependent Stochastic Travel Times

This record contains the preprint manuscript and LaTeX source files for “QR-BHDQN: Risk-Conditioned Distributional Hierarchical DQN for Multi-Depot Vehicle Routing with Time-Dependent Stochastic Travel Times.” The work studies risk-sensitive reinforcement learning for multi-depot vehicle routing under time-dependent stochastic travel times, with a focus on distributional value estimation, CVaR-based decision making, and hierarchical routing decisions. This version is a preprint prepared for academic communication and further journal revision.