Simultaneous Pickup and Delivery Traveling Salesman Problem considering the Express Lockers Using Attention Route...

Area of Science:

• Operations Research
• Artificial Intelligence
• Logistics Management

Background:

• Traditional Traveling Salesman Problem (TSP) models are insufficient for complex courier scenarios.
• Courier services face challenges in optimizing multi-trip routes with locker integration.
• Balancing costs like back order, lost sales, and travel time is crucial for efficiency.

Purpose of the Study:

• To develop a novel route design model for simultaneous pickup and delivery services incorporating express lockers.
• To minimize total operational costs, including back order, lost sale, and traveling time, while fulfilling all delivery requests.
• To leverage advanced AI techniques for optimizing courier path selection in multi-trip scenarios.

Main Methods:

• A modified deep Q-learning network with masked multi-head attention was designed.
• The model considers courier and express locker capacities for route optimization.
• Simultaneous pickup and delivery scenarios with multiple courier trips were analyzed.

Main Results:

• The proposed algorithm achieved superior optimal solutions compared to traditional stochastic optimization methods.
• The algorithm demonstrated efficient O(n) computational time during evaluation.
• Reinforcement learning proved more effective than stochastic methods, consuming less power and time.

Conclusions:

• The developed deep Q-learning model offers a more efficient and effective approach to courier route design.
• This AI-driven method is suitable for large-scale data and broad deployment in logistics.
• Integrating express lockers and multi-trip strategies significantly enhances delivery route optimization.