A GRASP-based Solution Construction Approach for the Multi-Vehicle Profitable Pickup and Delivery Problem

Abeer I Alhujaylan,Manar I

doi:10.14569/ijacsa.2019.0100412

Abstract

With the advancement of e-commerce and Internet shopping, the high competition between carriers has made many companies rethink their service mechanisms to customers, in order to ensure that they stay competitive in the market. Therefore, companies with limited resources focus on serving only customers who provide high profits at the lowest possible cost. The Multi-Vehicle Proﬁtable Pickup and Delivery Problem (MVPPDP) is a vehicle routing problem and one variant of the Selective Pickup and Delivery Problem (SPDP) that is considered to plan the services for these types of companies. The MVPPDP aims to serve only the profitable customers, where the products are transformed from a selection of pickup customers to the corresponding delivery customers, within a given travel time limit. In this paper, we utilize the construction phase of the well-known Greedy Randomized Adaptive Search Procedure (GRASP) to build initial solutions for the MVPPDP. The performance of the proposed method is compared with two greedy construction heuristics that were previously used in the literature to build the initial solutions of the MVPPDP. The results proved the effectiveness of the proposed method, where eight new initial solutions are obtained for the problem. Our approach is especially beneficial for building a population of solutions that combine both diversity and quality, which can help to obtain good solutions in the improvement phase of the problem.

Highlights

Transportation management is considered one of the most difficult problems facing people and governments in different countries all over the world
The stopping criterion is the number of iterations which is 100 for all datasets except the large-sized data instances which was set to 20 iterations only, due to time limitation
The algorithm uses the concept of a Restricted Candidate List (RCL) to combine between random and greedy properties, which can help in the diversification of the search, and is especially beneficial for population-based metaheuristics

Summary

Introduction

Transportation management is considered one of the most difficult problems facing people and governments in different countries all over the world. Millions of people use land, sea, or air transport means to commute from one place to another, raising the need to optimize the planning of these services, in order to reduce their cost as well as their negative environmental impacts. A lot of research has been conducted recently to address these problems in the fields of computer science, operations research, and industrial engineering. In particular, has received a great interest from researchers, due to its huge volume. Research efforts try to optimize the daily use of the means of transportation, such as cars, buses, trucks, trains, motorcycles, trams, etc. Among the most known land transport problems are: vehicle routing problems [1], pickup and delivery problems [2], bus scheduling problems [3], truck routing problems [4], cash transportation problems [5], railroad blocking problems [6], and others [7][8][9]

Methods

Results

Conclusion