Solving the ridesharing problem with Non-homogeneous vehicles by using an improved genetic algorithm and the social preferences of the users

Abstract

Most of the existing ridesharing systems perform travel planning based only on the two criteria of spatial and temporal similarity of the travelers. In general, neglecting the social preferences leads to reducing the users' willingness to use ridesharing services. To achieve this purpose, a system should be designed and implemented not just based on the two necessary conditions of spatial and temporal similarities, but also based on the similarities between the users in terms of their social and personal preferences to plan a travel. This study aims to create and implement a suitable model for ridesharing systems by using vehicles with different capacities and considering the users' social preferences using an advanced genetic algorithm. In this study, two innovative mutation operators and two local search algorithms have been applied to improve the genetic algorithm in this particular case. In this model, a mechanism has been designed to help the users to share their feedbacks on their travel experience on a hypothetical social network with other users. Then, the effect of the users' opinions on each other whether to use or not to use this system was analyzed and examined. As the users' interest in using this system increased, the results obtained from this analysis and the evaluation of the implemented model indicate efficiency and success of the system.