A genetic and insertion heuristic algorithm for solving the dynamic ridematching problem with time windows

Abstract

In this paper, we address the dynamic ridematching problem with time windows in dynamic ridesharing. The dynamic ridesharing is a special type of ridesharing where the participants form ridesharing on short notice. The ridematching problem is to assign riders to drivers and to define the ordering and timing of the riders' pickup and delivery. Because not all information is known in advance, the problem is dynamic. This is an optimization problem where we optimize a multicriteria objective function. We consider minimizing the total travel distance and time of the drivers and the total travel time of the riders and maximizing the number of the transported riders.We propose a genetic and insertion heuristic algorithm for solving the addressed problem. In the first stage, the algorithm works as a genetic algorithm while in the second stage it works as an insertion heuristic that modifies the solution of the genetic algorithm to do ridematching in real-time. In addition, we provide datasets for the ridematching problem, derived from realistic data, to test the algorithm. Experimentation results indicate that the algorithm can successfully solve the problem by providing answers in real-time and it can be easily tuned between response time and solution quality.