Session: 17-01-01: Research Posters

Paper Number: 149808

149808 - Last-Mile Delivery Optimization: A Case-Study Using Meituan Dataset 

Last-mile delivery has become an important field of interest in logistics due to the rapid increase in e-commerce. It refers to the final step in the logistics which aims at delivering goods from a distribution center to the end customer. Even though it focuses only on the final step, the time window constraints make it challenging for retailers to handle demands efficiently. For the delivery, several forms exist which include human-operated vehicles or autonomous vehicles such as robots and drones. Human-operated vehicles remain the most common mode for last-mile deliveries. Several research problems exist in optimizing the cost and efficiency of last-mile deliveries. However, most solutions do not consider the pre-book criteria which can significantly impact delivery scheduling and efficiency. In our research, we propose a solution to the last-mile delivery problem using a Capacity Vehicle Routing Problem with Time Windows (CVRPTW) framework. For this approach, we consider raw data from Meituan, a prominent food takeaway and delivery company in China. By incorporating penalties and pre-booking options, we employ the Gurobi optimizer to solve the problem using the Meituan dataset. In the initial problem setup, we consider ideal scenarios such as linear trajectories and the absence of obstacles. A subset of the Meituan dataset is analyzed which consists of one depot with five human-operated vehicles and 20 customers. Initial results indicate that pre-booking can improve overall customer satisfaction by allowing them to schedule deliveries in advance. Additionally, penalty settings prove crucial for the optimization problem as they can lead to higher travel times and costs if not managed correctly. The results aim to provide a comprehensive solution that addresses the complexities of last-mile delivery by integrating pre-booking options and optimizing delivery routes. The initial findings suggest potential improvements in efficiency, cost savings and customer satisfaction. Also, the findings laid the groundwork for performing testing and validation on real-world data. For future work, testing will be done using the proposed solution on real-time data. For this study, we will focus on the La Défense sector in the Paris region, one of the busiest districts with numerous enterprises and restaurants. This problem will be addressed by considering real-time traffic data, obstacles and non-linear trajectories. In order to enhance the adaptability and robustness of the optimization model alternative algorithms such as Artificial Neural Networks will be deployed. Exploring dynamic routing strategies and real-time data integration could further enhance the efficiency and responsiveness of last-mile delivery operations.

Presenting Author: Swaminath Venkateswaran Léonard de Vinci Pôle Universitaire, Research Center

Presenting Author Biography: Swaminath Venkateswaran obtained his Doctorate of Philosophy with honours in Mechanical Engineering from Ecole Centrale de Nantes, France in 2020. His research areas/expertise include Product design & analysis, Kinematics of mechanisms, Control of robots, Cobots for the circular economy and Industry 4.0. Currently, Swaminath works as an Assistant Professor at the Leonardo da Vinci engineering school (ESILV), Paris. His teaching activities are centred around the domain of Mechatronics and Industrial engineering for the Bachelors's & Master's levels. His research activities are affiliated with the group "New Materials, Intelligent Systems, and Innovative Companies" of the Da Vinci Research Center (DVRC).

Authors:

Hang Zhou Leonardo Da Vinci Engineering School (ESILV)
Zhe Yuan Léonard de Vinci Pôle Universitaire, Research Center
Swaminath Venkateswaran Léonard de Vinci Pôle Universitaire, Research Center