Attention based model for learning to solve the Heterogeneous Capacitated Vehicle Routing Problem (HCVRP) with both min-max and min-sum objective. Training with REINFORCE with greedy rollout baseline.

For more details, please see our paper： Jingwen Li, Yining Ma, Ruize Gao, Zhiguang Cao, Andrew Lim, Wen Song, Jie Zhang. Deep Reinforcement Learning for Solving the Heterogeneous Capacitated Vehicle Routing Problem. IEEE Transactions on Cybernetics, 2021. If this code is useful for your work, please cite our paper,

@article{li2021hcvrp,
  title={Deep Reinforcement Learning for Solving the Heterogeneous Capacitated Vehicle Routing Problem},
  author={Li, Jingwen and Ma, Yining and Gao, Ruize and Cao, Zhiguang and Andrew, Lim and Song, Wen and Zhang, Jie},
  journal={IEEE Transactions on Cybernetics},
  year={2021}
}

• Python>=3.7
• NumPy
• SciPy
• PyTorch=1.3.0
• tqdm
• tensorboard_logger
• Matplotlib (optional, only for plotting)

For more details, please see the fleet_v3 and fleet_v5 for HCVRP with three vehicles and five vehicles, respectively.