Luyan-Logistic-Optimization-Project

Research

• Developed and refined statistical models using cleaned data and multiple linear regression to accurately forecast delivery times and efficiency for a top-tier pharmacy company
• Implemented and compared various optimization techniques, including Google OR-Tools, PyVRP, VRPy, and advanced strategies from NeurIPS 2022 and Amazon, to enhance delivery efficiency and reduce costs
• Employed Web crawlers with Map API to capture real-world transit data and refine optimization models
• Crafted innovative graph algorithms to optimize routes, reduce fuel use, and elevate delivery efficiency

Conclusion

• PyVRP achieved a significantly lower total distance (706.758 km) compared to OR-Tools, suggesting it might have found a more efficient set of routes in terms of distance. However, it did so with fewer routes (12) and higher average demand per route.
• OR-Tools, on the other hand, resulted in a higher total distance (1138.509 km) but used more routes (17), vehicles (30), and less demand for each route. This could indicate an approach to balance route distances and the number of routes or vehicles used.

Members

• Haiyue Zhang, Shuo Wang, Yuhong Shao

Mentors

• Professor Frank Quan(UIUC)
• Litong Liu(Cornell)